Merge branch 'vad'

* vad:
update llm prompt
update whipser model default
update whipser model
Using local model for FunASR.
Track some large files or binary files by Git-LFS.
fix translate 7b bug
update prompt keywords
在FunASR模型初始化时添加disable_update参数以禁用更新。
change config folder structures
update and add keywords list
[fix]: update web.
add translate from 7b model when partial==False

.gitattributes

@@ -36,4 +36,8 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 assets/jfk.flac filter=lfs diff=lfs merge=lfs -text
 *.gguf filter=lfs diff=lfs merge=lfs -text
 moyoyo_asr_models/*.gguf filter=lfs diff=lfs merge=lfs -text
+moyoyo_asr_models/punc_ct-transformer_cn-en-common-vocab471067-large/jieba_usr_dict filter=lfs diff=lfs merge=lfs -text
+moyoyo_asr_models/punc_ct-transformer_cn-en-common-vocab471067-large/jieba.c.dict filter=lfs diff=lfs merge=lfs -text
+*.png filter=lfs diff=lfs merge=lfs -text
+*.wav filter=lfs diff=lfs merge=lfs -text
 *.icns filter=lfs diff=lfs merge=lfs -text

config/__init__.py

@@ -0,0 +1,2 @@
+from .settings import *
+from .prompt import *

{moyoyo_asr_models → config}/hotwords.json

@@ -3,5 +3,10 @@
     "GO SIM": "GOSIM",
     "go sim": "GOSIM",
     "GO SAME": "GOSIM",
-    "go same": "GOSIM"
+    "go same": "GOSIM",
+    "GoSync": "GOSIM",
+    "CSN": "CSDN",
+    "CSDF": "CSDN",
+    "CSTN": "CSDN",
+    "OpenAZI": "Open AGI"
 }

config/keywords.txt

@@ -0,0 +1,4 @@
+OpenAGI
+GOSIM
+Rust
+LLaMA Factory

config/prompt.py

@@ -0,0 +1,49 @@
+from .settings import CONFIG_DIR
+import json
+
+LLM_SYS_PROMPT_ZH = """
+你是一个中英文翻译专家，将用户输入的中文翻译成英文。对于非中文内容，它将提供中文翻译结果。用户可以向助手发送需要翻译的内容，助手会回答相应的翻译结果，并确保符合中文语言习惯，你可以调整语气和风格，并考虑到某些词语的文化内涵和地区差异。同时作为翻译家，需将原文翻译成具有信达雅标准的译文。"信" 即忠实于原文的内容与意图；"达" 意味着译文应通顺易懂，表达清晰；"雅" 则追求译文的文化审美和语言的优美。目标是创作出既忠于原作精神，又符合目标语言文化和读者审美的翻译。注意，翻译的文本只能包含拼音化字符，不能包含任何中文字符。
+"""
+
+LLM_SYS_PROMPT_EN = """
+你是一个英中文翻译专家，将用户输入的英文翻译成中文，用户可以向助手发送需要翻译的内容，助手会回答相应的翻译结果，并确保符合英文语言习惯，你可以调整语气和风格，并考虑到某些词语的文化内涵和地区差异。
+注意不要翻译人名和专业术语。
+同时作为翻译家，需将英文翻译成具有信达雅标准的中文。"信" 即忠实于原文的内容与意图；"达" 意味着译文应通顺易懂，表达清晰；"雅" 则追求译文的文化审美和语言的优美。目标是创作出既忠于原作精神，又符合目标语言文化和读者审美的翻译。
+"""
+
+hotwords_file = CONFIG_DIR / 'hotwords.txt'
+hotwords_json = json.loads((CONFIG_DIR / 'hotwords.json').read_text())
+
+# 翻译提示词
+keywords_list = [i.strip() for i in (CONFIG_DIR / 'keywords.txt').read_text().split('\n') if i.strip()]
+keywords_mapping_string = '\n'.join([f'  * {value}' for value in keywords_list ])
+
+LLM_SYS_7B_PROMPT_EN = """
+你是一个英中文翻译专家，尤其擅长将专业学术语言翻译成浅显易懂的口语。请你帮我将以下英文段落翻译成中文，风格与中文科普读物相似。
+你只需要翻译文本内容，不要做任何解释，也不要进行任何问答。
+
+规则：
+- 注意翻译时保留术语，例如 FLAC，JPEG 等。保留公司缩写，例如 Microsoft, Amazon, OpenAI 等；
+- 人物的名称不需要翻译；
+- 在翻译专业术语时，第一次出现时要在括号里面写上英文原文，例如：“生成式 AI (Generative AI)”，之后就可以只写中文了；
+- 以下是常见的AI相关术语，这部分的术语不需要翻译；
+
+{keywords_mapping_string}
+
+文本：
+""".format(keywords_mapping_string=keywords_mapping_string)
+
+LLM_SYS_7B_PROMPT_ZH = """
+你是一个中英文翻译专家，请将以下文本从中文翻译成英文，但保留所有英文专业术语、产品名称、代码片段和专有名词的原始英文形式。遇到英文专业术语或需要保留的内容时，请使用原始英文表达，不要翻译。
+翻译规则：
+1. 保留以下内容的原始英文形式，不翻译：
+   - 技术术语和专业词汇
+   - 产品名称、品牌名称
+   - 代码片段、函数名、变量名
+   - 专有名词、缩写和首字母缩略词
+2. 翻译其余内容时，请确保：
+   - 保持原文的段落结构
+   - 翻译内容符合中文表达习惯
+   - 保持专业性和准确性
+如遇到难以判断是否需要保留英文的情况，请优先保留原始英文形式。
+文本："""

config.py → config/settings.py

@@ -3,7 +3,6 @@ import re
 import logging
 import json
 
-
 DEBUG = False
 LOG_LEVEL = logging.DEBUG if DEBUG else logging.WARNING
 
@@ -14,9 +13,9 @@ logging.basicConfig(
     filename='translator.log',
     datefmt="%H:%M:%S"
 )
-# save pipelines data to disk
+
 SAVE_DATA_SAVE = False
-# Add terminal log
+
 console_handler = logging.StreamHandler()
 console_handler.setLevel(LOG_LEVEL)
 console_formatter = logging.Formatter("%(asctime)s - %(levelname)s - %(message)s")
@@ -28,9 +27,10 @@ FRAME_SCOPE_TIME_THRESHOLD = 4
 # 最长语音时长
 MAX_SPEECH_DURATION_S = 15
 
-BASE_DIR = pathlib.Path(__file__).parent
+BASE_DIR = pathlib.Path(__file__).parent.parent
 MODEL_DIR = BASE_DIR / "moyoyo_asr_models"
 ASSERT_DIR = BASE_DIR / "assets"
+CONFIG_DIR = BASE_DIR / "config"
 
 SAMPLE_RATE = 16000
 # 标点
@@ -54,25 +54,14 @@ MAX_LENGTH_ZH = 4
 WHISPER_PROMPT_EN = ""  # "The following is an English sentence."
 MAX_LENGTH_EN = 8
 
-WHISPER_MODEL_EN = 'medium-q5_0'
-# WHISPER_MODEL = 'large-v3-turbo-q5_0'
+# WHISPER_MODEL_EN = 'medium-q5_0'
+WHISPER_MODEL_EN = 'large-v3-turbo-q5_0'
 # WHISPER_MODEL_ZH = 'small'
 WHISPER_MODEL_ZH = 'large-v3-turbo-q5_0'
 # LLM
 LLM_MODEL_PATH = (MODEL_DIR / "qwen2.5-1.5b-instruct-q5_0.gguf").as_posix()
-LLM_LARGE_MODEL_PATH = (MODEL_DIR / "qwen2.5-1.5b-instruct-q5_0.gguf").as_posix()
-# LLM_LARGE_MODEL_PATH = (MODEL_DIR / "qwen2.5-7b-instruct-q5_0-00001-of-00002.gguf").as_posix()
+# LLM_LARGE_MODEL_PATH = (MODEL_DIR / "qwen2.5-1.5b-instruct-q5_0.gguf").as_posix()
+LLM_LARGE_MODEL_PATH = (MODEL_DIR / "qwen2.5-7b-instruct-q5_0-00001-of-00002.gguf").as_posix()
 
 # VAD
-VAD_MODEL_PATH = (MODEL_DIR / "silero-vad" / "silero_vad.onnx").as_posix()
-
-LLM_SYS_PROMPT_ZH = """
-你是一个中英文翻译专家，将用户输入的中文翻译成英文。对于非中文内容，它将提供中文翻译结果。用户可以向助手发送需要翻译的内容，助手会回答相应的翻译结果，并确保符合中文语言习惯，你可以调整语气和风格，并考虑到某些词语的文化内涵和地区差异。同时作为翻译家，需将原文翻译成具有信达雅标准的译文。"信" 即忠实于原文的内容与意图；"达" 意味着译文应通顺易懂，表达清晰；"雅" 则追求译文的文化审美和语言的优美。目标是创作出既忠于原作精神，又符合目标语言文化和读者审美的翻译。注意，翻译的文本只能包含拼音化字符，不能包含任何中文字符。
-"""
-
-LLM_SYS_PROMPT_EN = """
-你是一个英中文翻译专家，将用户输入的英文翻译成中文，用户可以向助手发送需要翻译的内容，助手会回答相应的翻译结果，并确保符合英文语言习惯，你可以调整语气和风格，并考虑到某些词语的文化内涵和地区差异。同时作为翻译家，需将英文翻译成具有信达雅标准的中文。"信" 即忠实于原文的内容与意图；"达" 意味着译文应通顺易懂，表达清晰；"雅" 则追求译文的文化审美和语言的优美。目标是创作出既忠于原作精神，又符合目标语言文化和读者审美的翻译。
-"""
-
-hotwords_file = MODEL_DIR / 'hotwords.txt'
-hotwords_json = json.loads((MODEL_DIR / 'hotwords.json').read_text())
+VAD_MODEL_PATH = (MODEL_DIR / "silero-vad" / "silero_vad.onnx").as_posix()

frontend/assets/{index-d03605fd.css → index-c1b225bf.css}

@@ -1 +1 @@
-html,body{width:100%;height:100%}input::-ms-clear,input::-ms-reveal{display:none}*,*:before,*:after{box-sizing:border-box}html{font-family:sans-serif;line-height:1.15;-webkit-text-size-adjust:100%;-ms-text-size-adjust:100%;-ms-overflow-style:scrollbar;-webkit-tap-highlight-color:rgba(0,0,0,0)}body{margin:0}[tabindex="-1"]:focus{outline:none}hr{box-sizing:content-box;height:0;overflow:visible}h1,h2,h3,h4,h5,h6{margin-top:0;margin-bottom:.5em;font-weight:500}p{margin-top:0;margin-bottom:1em}abbr[title],abbr[data-original-title]{-webkit-text-decoration:underline dotted;text-decoration:underline;text-decoration:underline dotted;border-bottom:0;cursor:help}address{margin-bottom:1em;font-style:normal;line-height:inherit}input[type=text],input[type=password],input[type=number],textarea{-webkit-appearance:none}ol,ul,dl{margin-top:0;margin-bottom:1em}ol ol,ul ul,ol ul,ul ol{margin-bottom:0}dt{font-weight:500}dd{margin-bottom:.5em;margin-left:0}blockquote{margin:0 0 1em}dfn{font-style:italic}b,strong{font-weight:bolder}small{font-size:80%}sub,sup{position:relative;font-size:75%;line-height:0;vertical-align:baseline}sub{bottom:-.25em}sup{top:-.5em}pre,code,kbd,samp{font-size:1em;font-family:SFMono-Regular,Consolas,Liberation Mono,Menlo,Courier,monospace}pre{margin-top:0;margin-bottom:1em;overflow:auto}figure{margin:0 0 1em}img{vertical-align:middle;border-style:none}a,area,button,[role=button],input:not([type=range]),label,select,summary,textarea{touch-action:manipulation}table{border-collapse:collapse}caption{padding-top:.75em;padding-bottom:.3em;text-align:left;caption-side:bottom}input,button,select,optgroup,textarea{margin:0;color:inherit;font-size:inherit;font-family:inherit;line-height:inherit}button,input{overflow:visible}button,select{text-transform:none}button,html [type=button],[type=reset],[type=submit]{-webkit-appearance:button}button::-moz-focus-inner,[type=button]::-moz-focus-inner,[type=reset]::-moz-focus-inner,[type=submit]::-moz-focus-inner{padding:0;border-style:none}input[type=radio],input[type=checkbox]{box-sizing:border-box;padding:0}input[type=date],input[type=time],input[type=datetime-local],input[type=month]{-webkit-appearance:listbox}textarea{overflow:auto;resize:vertical}fieldset{min-width:0;margin:0;padding:0;border:0}legend{display:block;width:100%;max-width:100%;margin-bottom:.5em;padding:0;color:inherit;font-size:1.5em;line-height:inherit;white-space:normal}progress{vertical-align:baseline}[type=number]::-webkit-inner-spin-button,[type=number]::-webkit-outer-spin-button{height:auto}[type=search]{outline-offset:-2px;-webkit-appearance:none}[type=search]::-webkit-search-cancel-button,[type=search]::-webkit-search-decoration{-webkit-appearance:none}::-webkit-file-upload-button{font:inherit;-webkit-appearance:button}output{display:inline-block}summary{display:list-item}template{display:none}[hidden]{display:none!important}mark{padding:.2em;background-color:#feffe6}:root{font-family:Inter,system-ui,Avenir,Helvetica,Arial,sans-serif;line-height:1.5;font-weight:400;color-scheme:light dark;color:#ffffffde;background-color:#242424;font-synthesis:none;text-rendering:optimizeLegibility;-webkit-font-smoothing:antialiased;-moz-osx-font-smoothing:grayscale;-webkit-text-size-adjust:100%}a{font-weight:500;color:#646cff;text-decoration:inherit}a:hover{color:#535bf2}body{margin:0;display:flex;place-items:center;min-width:320px;height:auto;min-height:auto;color:#333;background:#fff}h1{font-size:3.2em;line-height:1.1}button{border-radius:8px;border:1px solid transparent;padding:.6em 1.2em;font-size:1em;font-weight:500;font-family:inherit;background-color:#1a1a1a;cursor:pointer;transition:border-color .25s}.card{border-bottom:solid 2px lightgray;align-items:center;justify-content:center;margin-top:40px;display:flex;max-width:1024px;width:100%}.seg-title{margin:24px 0;font-size:20px;font-weight:500}.seg-co{width:1022px;text-align:left;border-left:solid 6px midnightblue;padding-left:8px;margin-left:2px;margin-top:36px;line-height:24px}#app{margin:0 auto;padding:0;text-align:center;width:100%}.ant-btn{padding:4px 12px}@media (prefers-color-scheme: light){:root{color:#213547;background-color:#fff}a:hover{color:#747bff}button{background-color:#f9f9f9}}.ant-card{background:#f5f6fa}.ant-card .ant-card-actions{background-color:#e8e8f8cc!important}.ant-popover{max-width:800px!important}.ant-form-item{background:transparent;margin-bottom:40px!important}.ant-form-item .ant-form-item-explain-error{color:#ff4d4f;text-align:left!important}.ant-form-item-label label{font-size:18px!important;color:#1a1a1a!important;font-weight:500!important}.ant-tooltip{max-width:1022px!important}.ant-page-header-heading{width:1022px!important}.highlight{background:ghostwhite}.content[data-v-178d5f9f]{background-color:#fff;max-width:1280px;min-height:720px;margin:0 auto;display:flex;flex-direction:column;align-items:center;justify-content:space-between}.not-found-wrapper[data-v-aef52a59]{height:calc(100vh - 104px)}.config-content[data-v-f56961b9]{width:420px;margin:12px}.config-content .config-block[data-v-f56961b9]{margin:12px;padding-bottom:12px}.view-wrapper[data-v-f56961b9]{width:100%;height:100%;background-color:#fff}.view-wrapper .wrapper-width-fixed[data-v-f56961b9]{width:1280px}.view-wrapper .wrapper-width-auto[data-v-f56961b9]{width:100vw}.view-wrapper .content-wrapper[data-v-f56961b9]{text-align:left;max-width:100vw;min-width:320px;margin-bottom:64px;min-height:calc(100vh - 438px)}.view-wrapper .content-wrapper .chat-box[data-v-f56961b9]{width:100%;height:54vh;border-radius:4px;padding:12px;color:#2e2f33;font-size:18px}.view-wrapper .content-wrapper .chat-box-placeholder[data-v-f56961b9]{width:100%;height:58vh;border-radius:4px;padding:12px;font-size:18px;color:#a4a6ac}.view-wrapper .content-wrapper .actions-box[data-v-f56961b9]{display:flex;align-items:center;justify-content:space-between;margin:0 24px;height:48px}.view-wrapper .content-wrapper .actions-box .left-actions[data-v-f56961b9]{display:flex;align-items:center;justify-content:space-between;width:288px}.view-wrapper .content-wrapper .trans-list[data-v-f56961b9]{overflow-y:auto;width:100%;height:58vh;scrollbar-width:none;-ms-overflow-style:none}.view-wrapper .content-wrapper .trans-list[data-v-f56961b9]::-webkit-scrollbar{display:none}.view-wrapper .content-wrapper .trans-list .node[data-v-f56961b9]{margin-bottom:36px;width:100%!important;transition:all .3s ease}.view-wrapper .content-wrapper .trans-list .node .trans-time[data-v-f56961b9]{font-size:14px;color:#c4c6cc}.view-wrapper .content-wrapper .trans-list .node .trans-font-size-16[data-v-f56961b9]{font-size:16px}.view-wrapper .content-wrapper .trans-list .node .trans-font-size-18[data-v-f56961b9]{font-size:18px}.view-wrapper .content-wrapper .trans-list .node .trans-font-size-20[data-v-f56961b9]{font-size:20px}.view-wrapper .content-wrapper .trans-list .node .trans-font-size-22[data-v-f56961b9]{font-size:22px}.view-wrapper .content-wrapper .trans-list .node .trans-font-size-24[data-v-f56961b9]{font-size:24px}.view-wrapper .content-wrapper .trans-list .node .trans-src-lang[data-v-f56961b9]{color:#909299;font-weight:500}.view-wrapper .content-wrapper .trans-list .node .trans-dst-lang[data-v-f56961b9]{color:#2e2f33;font-weight:600}.view-wrapper .content-wrapper .trans-list .current_node[data-v-f56961b9]{background-color:#f0f1f7;padding:4px 8px}@keyframes highlight-f56961b9{0%{background-color:transparent}50%{background-color:#fff1ce80}to{background-color:transparent}}@keyframes slideIn-f56961b9{0%{opacity:0;transform:translateY(10px)}to{opacity:1;transform:translateY(0)}}.content-wrapper[data-v-c39ab0d6]{text-align:left;max-width:800px;min-width:320px;margin-bottom:64px;min-height:calc(100vh - 438px)}.content-wrapper .content-box[data-v-c39ab0d6]{padding:24px;height:240px;background-color:#e8e8e8;border-radius:16px;width:50%;margin:48px auto;min-width:300px}.content-wrapper .video-box[data-v-c39ab0d6]{max-width:800px;min-width:320px;width:90vw;height:auto}
+html,body{width:100%;height:100%}input::-ms-clear,input::-ms-reveal{display:none}*,*:before,*:after{box-sizing:border-box}html{font-family:sans-serif;line-height:1.15;-webkit-text-size-adjust:100%;-ms-text-size-adjust:100%;-ms-overflow-style:scrollbar;-webkit-tap-highlight-color:rgba(0,0,0,0)}body{margin:0}[tabindex="-1"]:focus{outline:none}hr{box-sizing:content-box;height:0;overflow:visible}h1,h2,h3,h4,h5,h6{margin-top:0;margin-bottom:.5em;font-weight:500}p{margin-top:0;margin-bottom:1em}abbr[title],abbr[data-original-title]{-webkit-text-decoration:underline dotted;text-decoration:underline;text-decoration:underline dotted;border-bottom:0;cursor:help}address{margin-bottom:1em;font-style:normal;line-height:inherit}input[type=text],input[type=password],input[type=number],textarea{-webkit-appearance:none}ol,ul,dl{margin-top:0;margin-bottom:1em}ol ol,ul ul,ol ul,ul ol{margin-bottom:0}dt{font-weight:500}dd{margin-bottom:.5em;margin-left:0}blockquote{margin:0 0 1em}dfn{font-style:italic}b,strong{font-weight:bolder}small{font-size:80%}sub,sup{position:relative;font-size:75%;line-height:0;vertical-align:baseline}sub{bottom:-.25em}sup{top:-.5em}pre,code,kbd,samp{font-size:1em;font-family:SFMono-Regular,Consolas,Liberation Mono,Menlo,Courier,monospace}pre{margin-top:0;margin-bottom:1em;overflow:auto}figure{margin:0 0 1em}img{vertical-align:middle;border-style:none}a,area,button,[role=button],input:not([type=range]),label,select,summary,textarea{touch-action:manipulation}table{border-collapse:collapse}caption{padding-top:.75em;padding-bottom:.3em;text-align:left;caption-side:bottom}input,button,select,optgroup,textarea{margin:0;color:inherit;font-size:inherit;font-family:inherit;line-height:inherit}button,input{overflow:visible}button,select{text-transform:none}button,html [type=button],[type=reset],[type=submit]{-webkit-appearance:button}button::-moz-focus-inner,[type=button]::-moz-focus-inner,[type=reset]::-moz-focus-inner,[type=submit]::-moz-focus-inner{padding:0;border-style:none}input[type=radio],input[type=checkbox]{box-sizing:border-box;padding:0}input[type=date],input[type=time],input[type=datetime-local],input[type=month]{-webkit-appearance:listbox}textarea{overflow:auto;resize:vertical}fieldset{min-width:0;margin:0;padding:0;border:0}legend{display:block;width:100%;max-width:100%;margin-bottom:.5em;padding:0;color:inherit;font-size:1.5em;line-height:inherit;white-space:normal}progress{vertical-align:baseline}[type=number]::-webkit-inner-spin-button,[type=number]::-webkit-outer-spin-button{height:auto}[type=search]{outline-offset:-2px;-webkit-appearance:none}[type=search]::-webkit-search-cancel-button,[type=search]::-webkit-search-decoration{-webkit-appearance:none}::-webkit-file-upload-button{font:inherit;-webkit-appearance:button}output{display:inline-block}summary{display:list-item}template{display:none}[hidden]{display:none!important}mark{padding:.2em;background-color:#feffe6}:root{font-family:Inter,system-ui,Avenir,Helvetica,Arial,sans-serif;line-height:1.5;font-weight:400;color-scheme:light dark;color:#ffffffde;background-color:#242424;font-synthesis:none;text-rendering:optimizeLegibility;-webkit-font-smoothing:antialiased;-moz-osx-font-smoothing:grayscale;-webkit-text-size-adjust:100%}a{font-weight:500;color:#646cff;text-decoration:inherit}a:hover{color:#535bf2}body{margin:0;display:flex;place-items:center;min-width:320px;height:auto;min-height:auto;color:#333;background:#fff}h1{font-size:3.2em;line-height:1.1}button{border-radius:8px;border:1px solid transparent;padding:.6em 1.2em;font-size:1em;font-weight:500;font-family:inherit;background-color:#1a1a1a;cursor:pointer;transition:border-color .25s}.card{border-bottom:solid 2px lightgray;align-items:center;justify-content:center;margin-top:40px;display:flex;max-width:1024px;width:100%}.seg-title{margin:24px 0;font-size:20px;font-weight:500}.seg-co{width:1022px;text-align:left;border-left:solid 6px midnightblue;padding-left:8px;margin-left:2px;margin-top:36px;line-height:24px}#app{margin:0 auto;padding:0;text-align:center;width:100%}.ant-btn{padding:4px 12px}@media (prefers-color-scheme: light){:root{color:#213547;background-color:#fff}a:hover{color:#747bff}button{background-color:#f9f9f9}}.ant-card{background:#f5f6fa}.ant-card .ant-card-actions{background-color:#e8e8f8cc!important}.ant-popover{max-width:800px!important}.ant-form-item{background:transparent;margin-bottom:40px!important}.ant-form-item .ant-form-item-explain-error{color:#ff4d4f;text-align:left!important}.ant-form-item-label label{font-size:18px!important;color:#1a1a1a!important;font-weight:500!important}.ant-tooltip{max-width:1022px!important}.ant-page-header-heading{width:1022px!important}.highlight{background:ghostwhite}.content[data-v-178d5f9f]{background-color:#fff;max-width:1280px;min-height:720px;margin:0 auto;display:flex;flex-direction:column;align-items:center;justify-content:space-between}.not-found-wrapper[data-v-aef52a59]{height:calc(100vh - 104px)}.config-content[data-v-1bbbedd4]{width:420px;margin:12px}.config-content .config-block[data-v-1bbbedd4]{margin:12px;padding-bottom:12px}.view-wrapper[data-v-1bbbedd4]{width:100%;height:100%;background-color:#fff}.view-wrapper .wrapper-width-fixed[data-v-1bbbedd4]{width:1280px}.view-wrapper .wrapper-width-auto[data-v-1bbbedd4]{width:100vw}.view-wrapper .content-wrapper[data-v-1bbbedd4]{text-align:left;max-width:100vw;min-width:320px;margin-bottom:64px;min-height:calc(100vh - 438px)}.view-wrapper .content-wrapper .chat-box[data-v-1bbbedd4]{width:100%;height:54vh;border-radius:4px;padding:12px;color:#2e2f33;font-size:18px}.view-wrapper .content-wrapper .chat-box-placeholder[data-v-1bbbedd4]{width:100%;height:58vh;border-radius:4px;padding:12px;font-size:18px;color:#a4a6ac}.view-wrapper .content-wrapper .actions-box[data-v-1bbbedd4]{display:flex;align-items:center;justify-content:space-between;margin:0 24px;height:48px}.view-wrapper .content-wrapper .actions-box .left-actions[data-v-1bbbedd4]{display:flex;align-items:center;justify-content:space-between;width:332px}.view-wrapper .content-wrapper .actions-box .right-actions[data-v-1bbbedd4]{display:flex;align-items:center;justify-content:space-between;width:108px}.view-wrapper .content-wrapper .trans-list[data-v-1bbbedd4]{overflow-y:auto;width:100%;height:58vh;scrollbar-width:none;-ms-overflow-style:none}.view-wrapper .content-wrapper .trans-list[data-v-1bbbedd4]::-webkit-scrollbar{display:none}.view-wrapper .content-wrapper .trans-list .node[data-v-1bbbedd4]{margin-bottom:36px;width:100%!important;transition:all .3s ease}.view-wrapper .content-wrapper .trans-list .node .trans-time[data-v-1bbbedd4]{font-size:14px;color:#c4c6cc}.view-wrapper .content-wrapper .trans-list .node .trans-font-size-16[data-v-1bbbedd4]{font-size:16px}.view-wrapper .content-wrapper .trans-list .node .trans-font-size-18[data-v-1bbbedd4]{font-size:18px}.view-wrapper .content-wrapper .trans-list .node .trans-font-size-20[data-v-1bbbedd4]{font-size:20px}.view-wrapper .content-wrapper .trans-list .node .trans-font-size-22[data-v-1bbbedd4]{font-size:22px}.view-wrapper .content-wrapper .trans-list .node .trans-font-size-24[data-v-1bbbedd4]{font-size:24px}.view-wrapper .content-wrapper .trans-list .node .trans-src-lang[data-v-1bbbedd4]{color:#909299;font-weight:500}.view-wrapper .content-wrapper .trans-list .node .trans-dst-lang[data-v-1bbbedd4]{color:#2e2f33;font-weight:600}.view-wrapper .content-wrapper .trans-list .current_node[data-v-1bbbedd4]{background-color:#f0f1f7;padding:4px 8px}@keyframes highlight-1bbbedd4{0%{background-color:transparent}50%{background-color:#fff1ce80}to{background-color:transparent}}@keyframes slideIn-1bbbedd4{0%{opacity:0;transform:translateY(10px)}to{opacity:1;transform:translateY(0)}}.content-wrapper[data-v-c39ab0d6]{text-align:left;max-width:800px;min-width:320px;margin-bottom:64px;min-height:calc(100vh - 438px)}.content-wrapper .content-box[data-v-c39ab0d6]{padding:24px;height:240px;background-color:#e8e8e8;border-radius:16px;width:50%;margin:48px auto;min-width:300px}.content-wrapper .video-box[data-v-c39ab0d6]{max-width:800px;min-width:320px;width:90vw;height:auto}

frontend/index.html

@@ -5,8 +5,8 @@
     <link rel="icon" type="image/svg+xml" href="./favicon.ico" />
     <meta name="viewport" content="width=device-width, initial-scale=1.0" />
     <title>Translator</title>
-    <script type="module" crossorigin src="./assets/index-f0dbe2f4.js"></script>
-    <link rel="stylesheet" href="./assets/index-d03605fd.css">
+    <script type="module" crossorigin src="./assets/index-a619a94d.js"></script>
+    <link rel="stylesheet" href="./assets/index-c1b225bf.css">
   </head>
   <body>
     <div id="app"></div>

moyoyo_asr_models/punc_ct-transformer_cn-en-common-vocab471067-large/.mv

@@ -0,0 +1 @@
+Revision:master,CreatedAt:1707184148

moyoyo_asr_models/punc_ct-transformer_cn-en-common-vocab471067-large/README.md

@@ -0,0 +1,274 @@
+---
+tasks:
+- punctuation
+domain:
+- audio
+model-type:
+- Classification
+frameworks:
+- pytorch
+metrics:
+- f1_score
+license: Apache License 2.0
+language: 
+- cn
+tags:
+- FunASR
+- CT-Transformer
+- Alibaba
+- ICASSP 2020
+datasets:
+  train:
+  - 100M-samples online data
+  test:
+  - wikipedia data test
+  - 10000 industrial Mandarin sentences test
+widgets:
+  - model_revision: v2.0.4
+    task: punctuation
+    inputs:
+      - type: text
+        name: input
+        title: 文本
+    examples:
+      - name: 1
+        title: 示例1
+        inputs:
+          - name: input
+            data: 那今天的会就到这里吧 happy new year 明年见
+    inferencespec:
+      cpu: 1 #CPU数量
+      memory: 4096
+---
+
+# Controllable Time-delay Transformer模型介绍
+
+[//]: # (Controllable Time-delay Transformer 模型是一种端到端标点分类模型。)
+
+[//]: # (常规的Transformer会依赖很远的未来信息，导致长时间结果不固定。Controllable Time-delay Transformer 在效果无损的情况下，有效控制标点的延时。)
+
+# Highlights
+- 中文标点通用模型：可用于语音识别模型输出文本的标点预测，支持中英文输入。
+  - 基于[Paraformer-large长音频模型](https://www.modelscope.cn/models/iic/speech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404-pytorch/summary)场景的使用
+  - 基于[FunASR框架](https://github.com/alibaba-damo-academy/FunASR)，可进行ASR，VAD，标点的自由组合
+  - 基于纯文本输入的标点预测
+
+
+## <strong>[FunASR开源项目介绍](https://github.com/alibaba-damo-academy/FunASR)</strong>
+<strong>[FunASR](https://github.com/alibaba-damo-academy/FunASR)</strong>希望在语音识别的学术研究和工业应用之间架起一座桥梁。通过发布工业级语音识别模型的训练和微调，研究人员和开发人员可以更方便地进行语音识别模型的研究和生产，并推动语音识别生态的发展。让语音识别更有趣！
+
+[**github仓库**](https://github.com/alibaba-damo-academy/FunASR)
+| [**最新动态**](https://github.com/alibaba-damo-academy/FunASR#whats-new) 
+| [**环境安装**](https://github.com/alibaba-damo-academy/FunASR#installation)
+| [**服务部署**](https://www.funasr.com)
+| [**模型库**](https://github.com/alibaba-damo-academy/FunASR/tree/main/model_zoo)
+| [**联系我们**](https://github.com/alibaba-damo-academy/FunASR#contact)
+
+
+## 模型原理介绍
+
+Controllable Time-delay Transformer是达摩院语音团队提出的高效后处理框架中的标点模块。本项目为中文通用标点模型，模型可以被应用于文本类输入的标点预测，也可应用于语音识别结果的后处理步骤，协助语音识别模块输出具有可读性的文本结果。
+
+<p align="center">
+<img src="fig/struct.png" alt="Controllable Time-delay Transformer模型结构"  width="500" />
+
+Controllable Time-delay Transformer 模型结构如上图所示，由 Embedding、Encoder 和 Predictor 三部分组成。Embedding 是词向量叠加位置向量。Encoder可以采用不同的网络结构，例如self-attention，conformer，SAN-M等。Predictor 预测每个token后的标点类型。
+
+在模型的选择上采用了性能优越的Transformer模型。Transformer模型在获得良好性能的同时，由于模型自身序列化输入等特性，会给系统带来较大时延。常规的Transformer可以看到未来的全部信息，导致标点会依赖很远的未来信息。这会给用户带来一种标点一直在变化刷新，长时间结果不固定的不良感受。基于这一问题，我们创新性的提出了可控时延的Transformer模型（Controllable Time-Delay Transformer, CT-Transformer），在模型性能无损失的情况下，有效控制标点的延时。
+
+更详细的细节见：
+- 论文： [CONTROLLABLE TIME-DELAY TRANSFORMER FOR REAL-TIME PUNCTUATION PREDICTION AND DISFLUENCY DETECTION](https://arxiv.org/pdf/2003.01309.pdf)
+
+
+#### 基于ModelScope进行推理
+
+以下为三种支持格式及api调用方式参考如下范例：
+- text.scp文件路径，例如example/punc_example.txt，格式为： key + "\t" + value
+```sh
+cat example/punc_example.txt
+1       跨境河流是养育沿岸人民的生命之源
+2       从存储上来说仅仅是全景图片它就会是图片的四倍的容量
+3       那今天的会就到这里吧happy new year明年见
+```
+```python
+from modelscope.pipelines import pipeline
+from modelscope.utils.constant import Tasks
+
+inference_pipline = pipeline(
+    task=Tasks.punctuation,
+    model='iic/punc_ct-transformer_cn-en-common-vocab471067-large',
+    model_revision="v2.0.4")
+
+rec_result = inference_pipline('example/punc_example.txt')
+print(rec_result)
+```
+- text二进制数据，例如：用户直接从文件里读出bytes数据
+```python
+rec_result = inference_pipline('我们都是木头人不会讲话不会动')
+```
+- text文件url，例如：https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_text/punc_example.txt
+```python
+rec_result = inference_pipline('https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_text/punc_example.txt')
+```
+
+
+## 基于FunASR进行推理
+
+下面为快速上手教程，测试音频（[中文](https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/vad_example.wav)，[英文](https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_en.wav)）
+
+### 可执行命令行
+在命令行终端执行：
+
+```shell
+funasr +model=paraformer-zh +vad_model="fsmn-vad" +punc_model="ct-punc" +input=vad_example.wav
+```
+
+注：支持单条音频文件识别，也支持文件列表，列表为kaldi风格wav.scp：`wav_id   wav_path`
+
+### python示例
+#### 非实时语音识别
+```python
+from funasr import AutoModel
+# paraformer-zh is a multi-functional asr model
+# use vad, punc, spk or not as you need
+model = AutoModel(model="paraformer-zh", model_revision="v2.0.4",
+                  vad_model="fsmn-vad", vad_model_revision="v2.0.4",
+                  punc_model="ct-punc-c", punc_model_revision="v2.0.4",
+                  # spk_model="cam++", spk_model_revision="v2.0.2",
+                  )
+res = model.generate(input=f"{model.model_path}/example/asr_example.wav", 
+            batch_size_s=300, 
+            hotword='魔搭')
+print(res)
+```
+注：`model_hub`：表示模型仓库，`ms`为选择modelscope下载，`hf`为选择huggingface下载。
+
+#### 实时语音识别
+
+```python
+from funasr import AutoModel
+
+chunk_size = [0, 10, 5] #[0, 10, 5] 600ms, [0, 8, 4] 480ms
+encoder_chunk_look_back = 4 #number of chunks to lookback for encoder self-attention
+decoder_chunk_look_back = 1 #number of encoder chunks to lookback for decoder cross-attention
+
+model = AutoModel(model="paraformer-zh-streaming", model_revision="v2.0.4")
+
+import soundfile
+import os
+
+wav_file = os.path.join(model.model_path, "example/asr_example.wav")
+speech, sample_rate = soundfile.read(wav_file)
+chunk_stride = chunk_size[1] * 960 # 600ms
+
+cache = {}
+total_chunk_num = int(len((speech)-1)/chunk_stride+1)
+for i in range(total_chunk_num):
+    speech_chunk = speech[i*chunk_stride:(i+1)*chunk_stride]
+    is_final = i == total_chunk_num - 1
+    res = model.generate(input=speech_chunk, cache=cache, is_final=is_final, chunk_size=chunk_size, encoder_chunk_look_back=encoder_chunk_look_back, decoder_chunk_look_back=decoder_chunk_look_back)
+    print(res)
+```
+
+注：`chunk_size`为流式延时配置，`[0,10,5]`表示上屏实时出字粒度为`10*60=600ms`，未来信息为`5*60=300ms`。每次推理输入为`600ms`（采样点数为`16000*0.6=960`），输出为对应文字，最后一个语音片段输入需要设置`is_final=True`来强制输出最后一个字。
+
+#### 语音端点检测（非实时）
+```python
+from funasr import AutoModel
+
+model = AutoModel(model="fsmn-vad", model_revision="v2.0.4")
+
+wav_file = f"{model.model_path}/example/asr_example.wav"
+res = model.generate(input=wav_file)
+print(res)
+```
+
+#### 语音端点检测（实时）
+```python
+from funasr import AutoModel
+
+chunk_size = 200 # ms
+model = AutoModel(model="fsmn-vad", model_revision="v2.0.4")
+
+import soundfile
+
+wav_file = f"{model.model_path}/example/vad_example.wav"
+speech, sample_rate = soundfile.read(wav_file)
+chunk_stride = int(chunk_size * sample_rate / 1000)
+
+cache = {}
+total_chunk_num = int(len((speech)-1)/chunk_stride+1)
+for i in range(total_chunk_num):
+    speech_chunk = speech[i*chunk_stride:(i+1)*chunk_stride]
+    is_final = i == total_chunk_num - 1
+    res = model.generate(input=speech_chunk, cache=cache, is_final=is_final, chunk_size=chunk_size)
+    if len(res[0]["value"]):
+        print(res)
+```
+
+#### 标点恢复
+```python
+from funasr import AutoModel
+
+model = AutoModel(model="ct-punc", model_revision="v2.0.4")
+
+res = model.generate(input="那今天的会就到这里吧 happy new year 明年见")
+print(res)
+```
+
+#### 时间戳预测
+```python
+from funasr import AutoModel
+
+model = AutoModel(model="fa-zh", model_revision="v2.0.4")
+
+wav_file = f"{model.model_path}/example/asr_example.wav"
+text_file = f"{model.model_path}/example/text.txt"
+res = model.generate(input=(wav_file, text_file), data_type=("sound", "text"))
+print(res)
+```
+
+更多详细用法（[示例](https://github.com/alibaba-damo-academy/FunASR/tree/main/examples/industrial_data_pretraining)）
+
+
+## 微调
+
+详细用法（[示例](https://github.com/alibaba-damo-academy/FunASR/tree/main/examples/industrial_data_pretraining)）
+
+
+
+
+
+## Benchmark
+中文标点预测通用模型在自采集的通用领域业务场景数据上有良好效果。训练数据大约100M个sample，每个sample可能包含1句或多句。
+
+### 自采集数据（20000+ samples）
+
+| precision                            | recall                                | f1_score                              |
+|:------------------------------------:|:-------------------------------------:|:-------------------------------------:|
+| <div style="width: 150pt">56.0</div> | <div style="width: 150pt">62.5</div>  | <div style="width:i 150pt">58.8</div>  | 
+
+## 使用方式以及适用范围
+
+运行范围
+- 现阶段只能在Linux-x86_64运行，不支持Mac和Windows。
+
+使用方式
+- 直接推理：可以直接对输入文本进行计算，输出带有标点的目标文字。
+
+使用范围与目标场景
+- 适合对文本数据进行标点预测，文本长度不限。
+
+## 相关论文以及引用信息
+
+```BibTeX
+@inproceedings{chen2020controllable,
+  title={Controllable Time-Delay Transformer for Real-Time Punctuation Prediction and Disfluency Detection},
+  author={Chen, Qian and Chen, Mengzhe and Li, Bo and Wang, Wen},
+  booktitle={ICASSP 2020-2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
+  pages={8069--8073},
+  year={2020},
+  organization={IEEE}
+}
+```
+

moyoyo_asr_models/punc_ct-transformer_cn-en-common-vocab471067-large/config.yaml

@@ -0,0 +1,46 @@
+model: CTTransformer
+model_conf:
+    ignore_id: 0
+    embed_unit: 516
+    att_unit: 516
+    dropout_rate: 0.1
+    punc_list:
+        - <unk>
+        - _
+        - ，
+        - 。
+        - ？
+        - 、
+    punc_weight:
+        - 1.0
+        - 1.0
+        - 1.0
+        - 1.0
+        - 1.0
+        - 1.0
+    sentence_end_id: 3
+
+encoder: SANMEncoder
+encoder_conf:
+    input_size: 516
+    output_size: 516
+    attention_heads: 12
+    linear_units: 2048
+    num_blocks: 12
+    dropout_rate: 0.1
+    positional_dropout_rate: 0.1
+    attention_dropout_rate: 0.0
+    input_layer: pe
+    pos_enc_class: SinusoidalPositionEncoder
+    normalize_before: true
+    kernel_size: 11
+    sanm_shfit: 0
+    selfattention_layer_type: sanm
+    padding_idx: 0
+
+tokenizer: CharTokenizer
+tokenizer_conf:
+  unk_symbol: <unk>
+
+
+

moyoyo_asr_models/punc_ct-transformer_cn-en-common-vocab471067-large/configuration.json

@@ -0,0 +1,15 @@
+{
+  "framework": "pytorch",
+  "task" : "punctuation",
+  "model": {"type" : "funasr"},
+  "pipeline": {"type":"funasr-pipeline"},
+  "model_name_in_hub": {
+    "ms":"iic/punc_ct-transformer_cn-en-common-vocab471067-large", 
+    "hf":""},
+  "file_path_metas": {
+    "init_param":"model.pt", 
+    "config":"config.yaml",
+    "tokenizer_conf": {"token_list": "tokens.json", "jieba_usr_dict": "jieba_usr_dict"},
+    "jieba_usr_dict": "jieba_usr_dict"
+  }
+}

moyoyo_asr_models/punc_ct-transformer_cn-en-common-vocab471067-large/example/punc_example.txt

@@ -0,0 +1,3 @@
+1	跨境河流是养育沿岸人民的生命之源长期以来为帮助下游地区防灾减灾中方技术人员在上游地区极为恶劣的自然条件下克服巨大困难甚至冒着生命危险向印方提供汛期水文资料处理紧急事件中方重视印方在跨境河流问题上的关切愿意进一步完善双方联合工作机制凡是中方能做的我们都会去做而且会做得更好我请印度朋友们放心中国在上游的任何开发利用都会经过科学规划和论证兼顾上下游的利益
+2	从存储上来说仅仅是全景图片它就会是图片的四倍的容量然后全景的视频会是普通视频八倍的这个存储的容要求而三d的模型会是图片的十倍这都对我们今天运行在的云计算的平台存储的平台提出了更高的要求
+3	那今天的会就到这里吧 happy new year 明年见

moyoyo_asr_models/punc_ct-transformer_cn-en-common-vocab471067-large/jieba.c.dict

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eb0a68f8bfa65e6956d59e8f9652b49f5c91952273e24737bc9a8c5e23055221
+size 41536866

moyoyo_asr_models/punc_ct-transformer_cn-en-common-vocab471067-large/jieba_usr_dict

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:059aa2af152734d5fd011a4d69cbab171f62347074db0db5157c17b5649010f5
+size 11280857

moyoyo_asr_models/punc_ct-transformer_cn-en-common-vocab471067-large/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7176cae922a872e130e6b88aef9a1153581711baf79c9124c7c95be383cd6f81
+size 1125507622

moyoyo_asr_models/speech_fsmn_vad_zh-cn-16k-common-pytorch/.mv

@@ -0,0 +1 @@
+Revision:master,CreatedAt:1707184291

moyoyo_asr_models/speech_fsmn_vad_zh-cn-16k-common-pytorch/README.md

@@ -0,0 +1,296 @@
+---
+tasks:
+- voice-activity-detection
+domain:
+- audio
+model-type:
+- VAD model
+frameworks:
+- pytorch
+backbone:
+- fsmn
+metrics:
+- f1_score
+license: Apache License 2.0
+language: 
+- cn
+tags:
+- FunASR
+- FSMN
+- Alibaba
+- Online
+datasets:
+  train:
+  - 20,000 hour industrial Mandarin task
+  test:
+  - 20,000 hour industrial Mandarin task
+widgets:
+  - task: voice-activity-detection
+    model_revision: v2.0.4
+    inputs:
+      - type: audio
+        name: input
+        title: 音频
+    examples:
+      - name: 1
+        title: 示例1
+        inputs:
+          - name: input
+            data: git://example/vad_example.wav 
+    inferencespec:
+      cpu: 1 #CPU数量
+      memory: 4096
+---
+
+# FSMN-Monophone VAD 模型介绍
+
+[//]: # (FSMN-Monophone VAD 模型)
+
+## Highlight
+- 16k中文通用VAD模型：可用于检测长语音片段中有效语音的起止时间点。
+  - 基于[Paraformer-large长音频模型](https://www.modelscope.cn/models/damo/speech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404-pytorch/summary)场景的使用
+  - 基于[FunASR框架](https://github.com/alibaba-damo-academy/FunASR)，可进行ASR，VAD，[中文标点](https://www.modelscope.cn/models/damo/punc_ct-transformer_zh-cn-common-vocab272727-pytorch/summary)的自由组合
+  - 基于音频数据的有效语音片段起止时间点检测
+
+## <strong>[FunASR开源项目介绍](https://github.com/alibaba-damo-academy/FunASR)</strong>
+<strong>[FunASR](https://github.com/alibaba-damo-academy/FunASR)</strong>希望在语音识别的学术研究和工业应用之间架起一座桥梁。通过发布工业级语音识别模型的训练和微调，研究人员和开发人员可以更方便地进行语音识别模型的研究和生产，并推动语音识别生态的发展。让语音识别更有趣！
+
+[**github仓库**](https://github.com/alibaba-damo-academy/FunASR)
+| [**最新动态**](https://github.com/alibaba-damo-academy/FunASR#whats-new) 
+| [**环境安装**](https://github.com/alibaba-damo-academy/FunASR#installation)
+| [**服务部署**](https://www.funasr.com)
+| [**模型库**](https://github.com/alibaba-damo-academy/FunASR/tree/main/model_zoo)
+| [**联系我们**](https://github.com/alibaba-damo-academy/FunASR#contact)
+
+
+## 模型原理介绍
+
+FSMN-Monophone VAD是达摩院语音团队提出的高效语音端点检测模型，用于检测输入音频中有效语音的起止时间点信息，并将检测出来的有效音频片段输入识别引擎进行识别，减少无效语音带来的识别错误。
+
+<p align="center">
+<img src="fig/struct.png" alt="VAD模型结构"  width="500" />
+
+FSMN-Monophone VAD模型结构如上图所示：模型结构层面，FSMN模型结构建模时可考虑上下文信息，训练和推理速度快，且时延可控；同时根据VAD模型size以及低时延的要求，对FSMN的网络结构、右看帧数进行了适配。在建模单元层面，speech信息比较丰富，仅用单类来表征学习能力有限，我们将单一speech类升级为Monophone。建模单元细分，可以避免参数平均，抽象学习能力增强，区分性更好。
+
+## 基于ModelScope进行推理
+
+- 推理支持音频格式如下：
+  - wav文件路径，例如：data/test/audios/vad_example.wav
+  - wav文件url，例如：https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/vad_example.wav
+  - wav二进制数据，格式bytes，例如：用户直接从文件里读出bytes数据或者是麦克风录出bytes数据。
+  - 已解析的audio音频，例如：audio, rate = soundfile.read("vad_example_zh.wav")，类型为numpy.ndarray或者torch.Tensor。
+  - wav.scp文件，需符合如下要求：
+
+```sh
+cat wav.scp
+vad_example1  data/test/audios/vad_example1.wav
+vad_example2  data/test/audios/vad_example2.wav
+...
+```
+
+- 若输入格式wav文件url，api调用方式可参考如下范例：
+
+```python
+from modelscope.pipelines import pipeline
+from modelscope.utils.constant import Tasks
+
+inference_pipeline = pipeline(
+    task=Tasks.voice_activity_detection,
+    model='iic/speech_fsmn_vad_zh-cn-16k-common-pytorch',
+    model_revision="v2.0.4",
+)
+
+segments_result = inference_pipeline(input='https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/vad_example.wav')
+print(segments_result)
+```
+
+- 输入音频为pcm格式，调用api时需要传入音频采样率参数fs，例如：
+
+```python
+segments_result = inference_pipeline(input='https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/vad_example.pcm', fs=16000)
+```
+
+- 若输入格式为文件wav.scp(注：文件名需要以.scp结尾)，可添加 output_dir 参数将识别结果写入文件中，参考示例如下：
+
+```python
+inference_pipeline(input="wav.scp", output_dir='./output_dir')
+```
+识别结果输出路径结构如下：
+
+```sh
+tree output_dir/
+output_dir/
+└── 1best_recog
+    └── text
+
+1 directory, 1 files
+```
+text：VAD检测语音起止时间点结果文件（单位：ms）
+
+- 若输入音频为已解析的audio音频，api调用方式可参考如下范例：
+
+```python
+import soundfile
+
+waveform, sample_rate = soundfile.read("vad_example_zh.wav")
+segments_result = inference_pipeline(input=waveform)
+print(segments_result)
+```
+
+- VAD常用参数调整说明（参考：vad.yaml文件）：
+  - max_end_silence_time：尾部连续检测到多长时间静音进行尾点判停，参数范围500ms～6000ms，默认值800ms(该值过低容易出现语音提前截断的情况)。
+  - speech_noise_thres：speech的得分减去noise的得分大于此值则判断为speech，参数范围：（-1,1）
+    - 取值越趋于-1，噪音被误判定为语音的概率越大，FA越高
+    - 取值越趋于+1，语音被误判定为噪音的概率越大，Pmiss越高
+    - 通常情况下，该值会根据当前模型在长语音测试集上的效果取balance
+    
+
+
+
+## 基于FunASR进行推理
+
+下面为快速上手教程，测试音频（[中文](https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/vad_example.wav)，[英文](https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_en.wav)）
+
+### 可执行命令行
+在命令行终端执行：
+
+```shell
+funasr ++model=paraformer-zh ++vad_model="fsmn-vad" ++punc_model="ct-punc" ++input=vad_example.wav
+```
+
+注：支持单条音频文件识别，也支持文件列表，列表为kaldi风格wav.scp：`wav_id   wav_path`
+
+### python示例
+#### 非实时语音识别
+```python
+from funasr import AutoModel
+# paraformer-zh is a multi-functional asr model
+# use vad, punc, spk or not as you need
+model = AutoModel(model="paraformer-zh", model_revision="v2.0.4",
+                  vad_model="fsmn-vad", vad_model_revision="v2.0.4",
+                  punc_model="ct-punc-c", punc_model_revision="v2.0.4",
+                  # spk_model="cam++", spk_model_revision="v2.0.2",
+                  )
+res = model.generate(input=f"{model.model_path}/example/asr_example.wav", 
+            batch_size_s=300, 
+            hotword='魔搭')
+print(res)
+```
+注：`model_hub`：表示模型仓库，`ms`为选择modelscope下载，`hf`为选择huggingface下载。
+
+#### 实时语音识别
+
+```python
+from funasr import AutoModel
+
+chunk_size = [0, 10, 5] #[0, 10, 5] 600ms, [0, 8, 4] 480ms
+encoder_chunk_look_back = 4 #number of chunks to lookback for encoder self-attention
+decoder_chunk_look_back = 1 #number of encoder chunks to lookback for decoder cross-attention
+
+model = AutoModel(model="paraformer-zh-streaming", model_revision="v2.0.4")
+
+import soundfile
+import os
+
+wav_file = os.path.join(model.model_path, "example/asr_example.wav")
+speech, sample_rate = soundfile.read(wav_file)
+chunk_stride = chunk_size[1] * 960 # 600ms
+
+cache = {}
+total_chunk_num = int(len((speech)-1)/chunk_stride+1)
+for i in range(total_chunk_num):
+    speech_chunk = speech[i*chunk_stride:(i+1)*chunk_stride]
+    is_final = i == total_chunk_num - 1
+    res = model.generate(input=speech_chunk, cache=cache, is_final=is_final, chunk_size=chunk_size, encoder_chunk_look_back=encoder_chunk_look_back, decoder_chunk_look_back=decoder_chunk_look_back)
+    print(res)
+```
+
+注：`chunk_size`为流式延时配置，`[0,10,5]`表示上屏实时出字粒度为`10*60=600ms`，未来信息为`5*60=300ms`。每次推理输入为`600ms`（采样点数为`16000*0.6=960`），输出为对应文字，最后一个语音片段输入需要设置`is_final=True`来强制输出最后一个字。
+
+#### 语音端点检测（非实时）
+```python
+from funasr import AutoModel
+
+model = AutoModel(model="fsmn-vad", model_revision="v2.0.4")
+
+wav_file = f"{model.model_path}/example/asr_example.wav"
+res = model.generate(input=wav_file)
+print(res)
+```
+
+#### 语音端点检测（实时）
+```python
+from funasr import AutoModel
+
+chunk_size = 200 # ms
+model = AutoModel(model="fsmn-vad", model_revision="v2.0.4")
+
+import soundfile
+
+wav_file = f"{model.model_path}/example/vad_example.wav"
+speech, sample_rate = soundfile.read(wav_file)
+chunk_stride = int(chunk_size * sample_rate / 1000)
+
+cache = {}
+total_chunk_num = int(len((speech)-1)/chunk_stride+1)
+for i in range(total_chunk_num):
+    speech_chunk = speech[i*chunk_stride:(i+1)*chunk_stride]
+    is_final = i == total_chunk_num - 1
+    res = model.generate(input=speech_chunk, cache=cache, is_final=is_final, chunk_size=chunk_size)
+    if len(res[0]["value"]):
+        print(res)
+```
+
+#### 标点恢复
+```python
+from funasr import AutoModel
+
+model = AutoModel(model="ct-punc", model_revision="v2.0.4")
+
+res = model.generate(input="那今天的会就到这里吧 happy new year 明年见")
+print(res)
+```
+
+#### 时间戳预测
+```python
+from funasr import AutoModel
+
+model = AutoModel(model="fa-zh", model_revision="v2.0.4")
+
+wav_file = f"{model.model_path}/example/asr_example.wav"
+text_file = f"{model.model_path}/example/text.txt"
+res = model.generate(input=(wav_file, text_file), data_type=("sound", "text"))
+print(res)
+```
+
+更多详细用法��[示例](https://github.com/alibaba-damo-academy/FunASR/tree/main/examples/industrial_data_pretraining)）
+
+
+## 微调
+
+详细用法（[示例](https://github.com/alibaba-damo-academy/FunASR/tree/main/examples/industrial_data_pretraining)）
+
+
+
+
+
+## 使用方式以及适用范围
+
+运行范围
+- 支持Linux-x86_64、Mac和Windows运行。
+
+使用方式
+- 直接推理：可以直接对长语音数据进行计算，有效语音片段的起止时间点信息（单位：ms）。
+
+## 相关论文以及引用信息
+
+```BibTeX
+@inproceedings{zhang2018deep,
+  title={Deep-FSMN for large vocabulary continuous speech recognition},
+  author={Zhang, Shiliang and Lei, Ming and Yan, Zhijie and Dai, Lirong},
+  booktitle={2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
+  pages={5869--5873},
+  year={2018},
+  organization={IEEE}
+}
+```

moyoyo_asr_models/speech_fsmn_vad_zh-cn-16k-common-pytorch/am.mvn

@@ -0,0 +1,8 @@
+<Nnet>
+<Splice> 400 400
+[ 0 ]
+<AddShift> 400 400
+<LearnRateCoef> 0 [ -8.311879 -8.600912 -9.615928 -10.43595 -11.21292 -11.88333 -12.36243 -12.63706 -12.8818 -12.83066 -12.89103 -12.95666 -13.19763 -13.40598 -13.49113 -13.5546 -13.55639 -13.51915 -13.68284 -13.53289 -13.42107 -13.65519 -13.50713 -13.75251 -13.76715 -13.87408 -13.73109 -13.70412 -13.56073 -13.53488 -13.54895 -13.56228 -13.59408 -13.62047 -13.64198 -13.66109 -13.62669 -13.58297 -13.57387 -13.4739 -13.53063 -13.48348 -13.61047 -13.64716 -13.71546 -13.79184 -13.90614 -14.03098 -14.18205 -14.35881 -14.48419 -14.60172 -14.70591 -14.83362 -14.92122 -15.00622 -15.05122 -15.03119 -14.99028 -14.92302 -14.86927 -14.82691 -14.7972 -14.76909 -14.71356 -14.61277 -14.51696 -14.42252 -14.36405 -14.30451 -14.23161 -14.19851 -14.16633 -14.15649 -14.10504 -13.99518 -13.79562 -13.3996 -12.7767 -11.71208 -8.311879 -8.600912 -9.615928 -10.43595 -11.21292 -11.88333 -12.36243 -12.63706 -12.8818 -12.83066 -12.89103 -12.95666 -13.19763 -13.40598 -13.49113 -13.5546 -13.55639 -13.51915 -13.68284 -13.53289 -13.42107 -13.65519 -13.50713 -13.75251 -13.76715 -13.87408 -13.73109 -13.70412 -13.56073 -13.53488 -13.54895 -13.56228 -13.59408 -13.62047 -13.64198 -13.66109 -13.62669 -13.58297 -13.57387 -13.4739 -13.53063 -13.48348 -13.61047 -13.64716 -13.71546 -13.79184 -13.90614 -14.03098 -14.18205 -14.35881 -14.48419 -14.60172 -14.70591 -14.83362 -14.92122 -15.00622 -15.05122 -15.03119 -14.99028 -14.92302 -14.86927 -14.82691 -14.7972 -14.76909 -14.71356 -14.61277 -14.51696 -14.42252 -14.36405 -14.30451 -14.23161 -14.19851 -14.16633 -14.15649 -14.10504 -13.99518 -13.79562 -13.3996 -12.7767 -11.71208 -8.311879 -8.600912 -9.615928 -10.43595 -11.21292 -11.88333 -12.36243 -12.63706 -12.8818 -12.83066 -12.89103 -12.95666 -13.19763 -13.40598 -13.49113 -13.5546 -13.55639 -13.51915 -13.68284 -13.53289 -13.42107 -13.65519 -13.50713 -13.75251 -13.76715 -13.87408 -13.73109 -13.70412 -13.56073 -13.53488 -13.54895 -13.56228 -13.59408 -13.62047 -13.64198 -13.66109 -13.62669 -13.58297 -13.57387 -13.4739 -13.53063 -13.48348 -13.61047 -13.64716 -13.71546 -13.79184 -13.90614 -14.03098 -14.18205 -14.35881 -14.48419 -14.60172 -14.70591 -14.83362 -14.92122 -15.00622 -15.05122 -15.03119 -14.99028 -14.92302 -14.86927 -14.82691 -14.7972 -14.76909 -14.71356 -14.61277 -14.51696 -14.42252 -14.36405 -14.30451 -14.23161 -14.19851 -14.16633 -14.15649 -14.10504 -13.99518 -13.79562 -13.3996 -12.7767 -11.71208 -8.311879 -8.600912 -9.615928 -10.43595 -11.21292 -11.88333 -12.36243 -12.63706 -12.8818 -12.83066 -12.89103 -12.95666 -13.19763 -13.40598 -13.49113 -13.5546 -13.55639 -13.51915 -13.68284 -13.53289 -13.42107 -13.65519 -13.50713 -13.75251 -13.76715 -13.87408 -13.73109 -13.70412 -13.56073 -13.53488 -13.54895 -13.56228 -13.59408 -13.62047 -13.64198 -13.66109 -13.62669 -13.58297 -13.57387 -13.4739 -13.53063 -13.48348 -13.61047 -13.64716 -13.71546 -13.79184 -13.90614 -14.03098 -14.18205 -14.35881 -14.48419 -14.60172 -14.70591 -14.83362 -14.92122 -15.00622 -15.05122 -15.03119 -14.99028 -14.92302 -14.86927 -14.82691 -14.7972 -14.76909 -14.71356 -14.61277 -14.51696 -14.42252 -14.36405 -14.30451 -14.23161 -14.19851 -14.16633 -14.15649 -14.10504 -13.99518 -13.79562 -13.3996 -12.7767 -11.71208 -8.311879 -8.600912 -9.615928 -10.43595 -11.21292 -11.88333 -12.36243 -12.63706 -12.8818 -12.83066 -12.89103 -12.95666 -13.19763 -13.40598 -13.49113 -13.5546 -13.55639 -13.51915 -13.68284 -13.53289 -13.42107 -13.65519 -13.50713 -13.75251 -13.76715 -13.87408 -13.73109 -13.70412 -13.56073 -13.53488 -13.54895 -13.56228 -13.59408 -13.62047 -13.64198 -13.66109 -13.62669 -13.58297 -13.57387 -13.4739 -13.53063 -13.48348 -13.61047 -13.64716 -13.71546 -13.79184 -13.90614 -14.03098 -14.18205 -14.35881 -14.48419 -14.60172 -14.70591 -14.83362 -14.92122 -15.00622 -15.05122 -15.03119 -14.99028 -14.92302 -14.86927 -14.82691 -14.7972 -14.76909 -14.71356 -14.61277 -14.51696 -14.42252 -14.36405 -14.30451 -14.23161 -14.19851 -14.16633 -14.15649 -14.10504 -13.99518 -13.79562 -13.3996 -12.7767 -11.71208 ]
+<Rescale> 400 400
+<LearnRateCoef> 0 [ 0.155775 0.154484 0.1527379 0.1518718 0.1506028 0.1489256 0.147067 0.1447061 0.1436307 0.1443568 0.1451849 0.1455157 0.1452821 0.1445717 0.1439195 0.1435867 0.1436018 0.1438781 0.1442086 0.1448844 0.1454756 0.145663 0.146268 0.1467386 0.1472724 0.147664 0.1480913 0.1483739 0.1488841 0.1493636 0.1497088 0.1500379 0.1502916 0.1505389 0.1506787 0.1507102 0.1505992 0.1505445 0.1505938 0.1508133 0.1509569 0.1512396 0.1514625 0.1516195 0.1516156 0.1515561 0.1514966 0.1513976 0.1512612 0.151076 0.1510596 0.1510431 0.151077 0.1511168 0.1511917 0.151023 0.1508045 0.1505885 0.1503493 0.1502373 0.1501726 0.1500762 0.1500065 0.1499782 0.150057 0.1502658 0.150469 0.1505335 0.1505505 0.1505328 0.1504275 0.1502438 0.1499674 0.1497118 0.1494661 0.1493102 0.1493681 0.1495501 0.1499738 0.1509654 0.155775 0.154484 0.1527379 0.1518718 0.1506028 0.1489256 0.147067 0.1447061 0.1436307 0.1443568 0.1451849 0.1455157 0.1452821 0.1445717 0.1439195 0.1435867 0.1436018 0.1438781 0.1442086 0.1448844 0.1454756 0.145663 0.146268 0.1467386 0.1472724 0.147664 0.1480913 0.1483739 0.1488841 0.1493636 0.1497088 0.1500379 0.1502916 0.1505389 0.1506787 0.1507102 0.1505992 0.1505445 0.1505938 0.1508133 0.1509569 0.1512396 0.1514625 0.1516195 0.1516156 0.1515561 0.1514966 0.1513976 0.1512612 0.151076 0.1510596 0.1510431 0.151077 0.1511168 0.1511917 0.151023 0.1508045 0.1505885 0.1503493 0.1502373 0.1501726 0.1500762 0.1500065 0.1499782 0.150057 0.1502658 0.150469 0.1505335 0.1505505 0.1505328 0.1504275 0.1502438 0.1499674 0.1497118 0.1494661 0.1493102 0.1493681 0.1495501 0.1499738 0.1509654 0.155775 0.154484 0.1527379 0.1518718 0.1506028 0.1489256 0.147067 0.1447061 0.1436307 0.1443568 0.1451849 0.1455157 0.1452821 0.1445717 0.1439195 0.1435867 0.1436018 0.1438781 0.1442086 0.1448844 0.1454756 0.145663 0.146268 0.1467386 0.1472724 0.147664 0.1480913 0.1483739 0.1488841 0.1493636 0.1497088 0.1500379 0.1502916 0.1505389 0.1506787 0.1507102 0.1505992 0.1505445 0.1505938 0.1508133 0.1509569 0.1512396 0.1514625 0.1516195 0.1516156 0.1515561 0.1514966 0.1513976 0.1512612 0.151076 0.1510596 0.1510431 0.151077 0.1511168 0.1511917 0.151023 0.1508045 0.1505885 0.1503493 0.1502373 0.1501726 0.1500762 0.1500065 0.1499782 0.150057 0.1502658 0.150469 0.1505335 0.1505505 0.1505328 0.1504275 0.1502438 0.1499674 0.1497118 0.1494661 0.1493102 0.1493681 0.1495501 0.1499738 0.1509654 0.155775 0.154484 0.1527379 0.1518718 0.1506028 0.1489256 0.147067 0.1447061 0.1436307 0.1443568 0.1451849 0.1455157 0.1452821 0.1445717 0.1439195 0.1435867 0.1436018 0.1438781 0.1442086 0.1448844 0.1454756 0.145663 0.146268 0.1467386 0.1472724 0.147664 0.1480913 0.1483739 0.1488841 0.1493636 0.1497088 0.1500379 0.1502916 0.1505389 0.1506787 0.1507102 0.1505992 0.1505445 0.1505938 0.1508133 0.1509569 0.1512396 0.1514625 0.1516195 0.1516156 0.1515561 0.1514966 0.1513976 0.1512612 0.151076 0.1510596 0.1510431 0.151077 0.1511168 0.1511917 0.151023 0.1508045 0.1505885 0.1503493 0.1502373 0.1501726 0.1500762 0.1500065 0.1499782 0.150057 0.1502658 0.150469 0.1505335 0.1505505 0.1505328 0.1504275 0.1502438 0.1499674 0.1497118 0.1494661 0.1493102 0.1493681 0.1495501 0.1499738 0.1509654 0.155775 0.154484 0.1527379 0.1518718 0.1506028 0.1489256 0.147067 0.1447061 0.1436307 0.1443568 0.1451849 0.1455157 0.1452821 0.1445717 0.1439195 0.1435867 0.1436018 0.1438781 0.1442086 0.1448844 0.1454756 0.145663 0.146268 0.1467386 0.1472724 0.147664 0.1480913 0.1483739 0.1488841 0.1493636 0.1497088 0.1500379 0.1502916 0.1505389 0.1506787 0.1507102 0.1505992 0.1505445 0.1505938 0.1508133 0.1509569 0.1512396 0.1514625 0.1516195 0.1516156 0.1515561 0.1514966 0.1513976 0.1512612 0.151076 0.1510596 0.1510431 0.151077 0.1511168 0.1511917 0.151023 0.1508045 0.1505885 0.1503493 0.1502373 0.1501726 0.1500762 0.1500065 0.1499782 0.150057 0.1502658 0.150469 0.1505335 0.1505505 0.1505328 0.1504275 0.1502438 0.1499674 0.1497118 0.1494661 0.1493102 0.1493681 0.1495501 0.1499738 0.1509654 ]
+</Nnet>

moyoyo_asr_models/speech_fsmn_vad_zh-cn-16k-common-pytorch/config.yaml

@@ -0,0 +1,56 @@
+frontend: WavFrontendOnline
+frontend_conf:
+    fs: 16000
+    window: hamming
+    n_mels: 80
+    frame_length: 25
+    frame_shift: 10
+    dither: 0.0
+    lfr_m: 5
+    lfr_n: 1
+
+model: FsmnVADStreaming
+model_conf:
+    sample_rate: 16000
+    detect_mode: 1 
+    snr_mode: 0
+    max_end_silence_time: 800
+    max_start_silence_time: 3000
+    do_start_point_detection: True
+    do_end_point_detection: True
+    window_size_ms: 200
+    sil_to_speech_time_thres: 150
+    speech_to_sil_time_thres: 150
+    speech_2_noise_ratio: 1.0
+    do_extend: 1
+    lookback_time_start_point: 200
+    lookahead_time_end_point: 100
+    max_single_segment_time: 60000
+    snr_thres: -100.0
+    noise_frame_num_used_for_snr: 100
+    decibel_thres: -100.0
+    speech_noise_thres: 0.6
+    fe_prior_thres: 0.0001
+    silence_pdf_num: 1
+    sil_pdf_ids: [0]
+    speech_noise_thresh_low: -0.1
+    speech_noise_thresh_high: 0.3
+    output_frame_probs: False
+    frame_in_ms: 10
+    frame_length_ms: 25
+    
+encoder: FSMN
+encoder_conf:
+    input_dim: 400
+    input_affine_dim: 140
+    fsmn_layers: 4
+    linear_dim: 250
+    proj_dim: 128
+    lorder: 20
+    rorder: 0
+    lstride: 1
+    rstride: 0
+    output_affine_dim: 140
+    output_dim: 248
+
+

moyoyo_asr_models/speech_fsmn_vad_zh-cn-16k-common-pytorch/configuration.json

@@ -0,0 +1,13 @@
+{
+  "framework": "pytorch",
+  "task" : "voice-activity-detection",
+  "pipeline": {"type":"funasr-pipeline"},
+  "model": {"type" : "funasr"},
+  "file_path_metas": {
+    "init_param":"model.pt", 
+    "config":"config.yaml",
+    "frontend_conf":{"cmvn_file": "am.mvn"}},
+  "model_name_in_hub": {
+    "ms":"iic/speech_fsmn_vad_zh-cn-16k-common-pytorch", 
+    "hf":""}
+}

moyoyo_asr_models/speech_fsmn_vad_zh-cn-16k-common-pytorch/example/vad_example.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a7431f0169ef76ef630c945a1d2c3675d8c8c2df2ae4a6b16f8a88ba1bccfbbb
+size 2261722

moyoyo_asr_models/speech_fsmn_vad_zh-cn-16k-common-pytorch/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3be75be477f0780277f3bae0fe489f48718f585f3a6e45d7dd1fbb1a4255fc5
+size 1721366

moyoyo_asr_models/speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/.mv

@@ -0,0 +1 @@
+Revision:master,CreatedAt:1727670560

moyoyo_asr_models/speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/README.md

@@ -0,0 +1,357 @@
+---
+tasks:
+- auto-speech-recognition
+domain:
+- audio
+model-type:
+- Non-autoregressive
+frameworks:
+- pytorch
+backbone:
+- transformer/conformer
+metrics:
+- CER
+license: Apache License 2.0
+language: 
+- cn
+tags:
+- FunASR
+- Paraformer
+- Alibaba
+- ICASSP2024
+- Hotword
+datasets:
+  train:
+  - 50,000 hour industrial Mandarin task
+  test:
+  - AISHELL-1-hotword dev/test
+indexing:
+   results:
+   - task:
+       name: Automatic Speech Recognition
+     dataset:
+       name: 50,000 hour industrial Mandarin task
+       type: audio    # optional
+       args: 16k sampling rate, 8404 characters  # optional
+     metrics:
+       - type: CER
+         value: 8.53%  # float
+         description: greedy search, withou lm, avg.
+         args: default
+       - type: RTF
+         value: 0.0251  # float
+         description: GPU inference on V100
+         args: batch_size=1
+widgets:
+  - task: auto-speech-recognition
+    inputs:
+      - type: audio
+        name: input
+        title: 音频
+    parameters:
+      - name: hotword
+        title: 热词
+        type: string
+    examples:
+      - name: 1
+        title: 示例1
+        inputs:
+          - name: input
+            data: git://example/asr_example.wav
+        parameters:
+          - name: hotword
+            value: 魔搭
+    model_revision: v2.0.4
+    inferencespec:
+      cpu: 8 #CPU数量
+      memory: 4096
+---
+
+# Paraformer-large模型介绍
+
+## Highlights
+Paraformer-large热词版模型支持热词定制功能：实现热词定制化功能，基于提供的热词列表进行激励增强，提升热词的召回率和准确率。
+
+
+## <strong>[FunASR开源项目介绍](https://github.com/alibaba-damo-academy/FunASR)</strong>
+<strong>[FunASR](https://github.com/alibaba-damo-academy/FunASR)</strong>希望在语音识别的学术研究和工业应用之间架起一座桥梁。通过发布工业级语音识别模型的训练和微调，研究人员和开发人员可以更方便地进行语音识别模型的研究和生产，并推动语音识别生态的发展。让语音识别更有趣！
+
+[**github仓库**](https://github.com/alibaba-damo-academy/FunASR) 
+|  [**最新动态**](https://github.com/alibaba-damo-academy/FunASR#whats-new) 
+| [**环境安装**](https://github.com/alibaba-damo-academy/FunASR#installation)
+| [**服务部署**](https://www.funasr.com)
+| [**模型库**](https://github.com/alibaba-damo-academy/FunASR/tree/main/model_zoo)
+| [**联系我们**](https://github.com/alibaba-damo-academy/FunASR#contact)
+
+
+## 模型原理介绍
+
+SeACoParaformer是阿里巴巴语音实验室提出的新一代热词定制化非自回归语音识别模型。相比于上一代基于CLAS的热词定制化方案，SeACoParaformer解耦了热词模块与ASR模型，通过后验概率融合的方式进行热词激励，使激励过程可见可控，并且热词召回率显著提升。
+
+<p align="center">
+<img src="fig/seaco.png" alt="SeACoParaformer模型结构"  width="380" />
+
+
+SeACoParaformer的模型结构与训练流程如上图所示，通过引入bias encoder进行热词embedding提取，bias decoder进行注意力建模，SeACoParaformer能够捕捉到Predictor输出和Decoder输出的信息与热词的相关性，并且预测与ASR结果同步的热词输出。通过后验概率的融合，实现热词激励。与ContextualParaformer相比，SeACoParaformer有明显的效果提升，如下图所示：
+
+<p align="center">
+<img src="fig/res.png" alt="SeACoParaformer模型结构"  width="700" />
+
+更详细的细节见：
+- 论文： [SeACo-Paraformer: A Non-Autoregressive ASR System with Flexible and Effective Hotword Customization Ability](https://arxiv.org/abs/2308.03266)
+
+## 复现论文中的结果
+```python
+from funasr import AutoModel
+
+model = AutoModel(model="iic/speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch",
+                  model_revision="v2.0.4",
+                  # vad_model="damo/speech_fsmn_vad_zh-cn-16k-common-pytorch",
+                  # vad_model_revision="v2.0.4",
+                  # punc_model="damo/punc_ct-transformer_zh-cn-common-vocab272727-pytorch",
+                  # punc_model_revision="v2.0.4",
+                  # spk_model="damo/speech_campplus_sv_zh-cn_16k-common",
+                  # spk_model_revision="v2.0.2",
+                  device="cuda:0"
+                  )
+
+res = model.generate(input="YOUR_PATH/aishell1_hotword_dev.scp",
+                     hotword='./data/dev/hotword.txt',
+                     batch_size_s=300,
+                    )
+fout1 = open("dev.output", 'w')
+for resi in res:
+    fout1.write("{}\t{}\n".format(resi['key'], resi['text']))
+
+res = model.generate(input="YOUR_PATH/aishell1_hotword_test.scp",
+                     hotword='./data/test/hotword.txt',
+                     batch_size_s=300,
+                    )
+fout2 = open("test.output", 'w')
+for resi in res:
+    fout2.write("{}\t{}\n".format(resi['key'], resi['text']))
+```
+
+## 基于ModelScope进行推理
+
+- 推理支��音频格式如下：
+  - wav文件路径，例如：data/test/audios/asr_example.wav
+  - pcm文件路径，例如：data/test/audios/asr_example.pcm
+  - wav文件url，例如：https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_zh.wav
+  - wav二进制数据，格式bytes，例如：用户直接从文件里读出bytes数据或者是麦克风录出bytes数据。
+  - 已解析的audio音频，例如：audio, rate = soundfile.read("asr_example_zh.wav")，类型为numpy.ndarray或者torch.Tensor。
+  - wav.scp文件，需符合如下要求：
+
+```sh
+cat wav.scp
+asr_example1  data/test/audios/asr_example1.wav
+asr_example2  data/test/audios/asr_example2.wav
+...
+```
+
+- 若输入格式wav文件url，api调用方式可参考如下范例：
+
+```python
+from modelscope.pipelines import pipeline
+from modelscope.utils.constant import Tasks
+
+inference_pipeline = pipeline(
+    task=Tasks.auto_speech_recognition,
+    model='iic/speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch', model_revision="v2.0.4")
+
+rec_result = inference_pipeline('https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_zh.wav', hotword='达摩院 魔搭')
+print(rec_result)
+```
+
+- 输入音频为pcm格式，调用api时需要传入音频采样率参数audio_fs，例如：
+
+```python
+rec_result = inference_pipeline('https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_zh.pcm', fs=16000, hotword='达摩院 魔搭')
+```
+
+- 输入音频为wav格式，api调用方式可参考如下范例:
+
+```python
+rec_result = inference_pipeline('asr_example_zh.wav', hotword='达摩院 魔搭')
+```
+
+- 若输入格式为文件wav.scp(注：文件名需要以.scp结尾)，可添加 output_dir 参数将识别结果写入文件中，api调用方式可参考如下范例:
+
+```python
+inference_pipeline("wav.scp", output_dir='./output_dir', hotword='达摩院 魔搭')
+```
+识别结果输出路径结构如下：
+
+```sh
+tree output_dir/
+output_dir/
+└── 1best_recog
+    ├── score
+    └── text
+
+1 directory, 3 files
+```
+
+score：识别路径得分
+
+text：语音识别结果文件
+
+
+- 若输入音频为已解析的audio音频，api调用方式可参考如下范例：
+
+```python
+import soundfile
+
+waveform, sample_rate = soundfile.read("asr_example_zh.wav")
+rec_result = inference_pipeline(waveform, hotword='达摩院 魔搭')
+```
+
+- ASR、VAD、PUNC模型自由组合
+
+可根据使用需求对VAD和PUNC标点模型进行自由组合，使用方式如下：
+```python
+inference_pipeline = pipeline(
+    task=Tasks.auto_speech_recognition,
+    model='iic/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch', model_revision="v2.0.4",
+    vad_model='iic/speech_fsmn_vad_zh-cn-16k-common-pytorch', vad_model_revision="v2.0.4",
+    punc_model='iic/punc_ct-transformer_zh-cn-common-vocab272727-pytorch', punc_model_revision="v2.0.3",
+    # spk_model="iic/speech_campplus_sv_zh-cn_16k-common",
+    # spk_model_revision="v2.0.2",
+)
+```
+若不使用PUNC模型，可配置punc_model=None，或不传入punc_model参数，如需加入LM模型，可增加配置lm_model='iic/speech_transformer_lm_zh-cn-common-vocab8404-pytorch'，并设置lm_weight和beam_size参数。
+
+## 基于FunASR进行推理
+
+下面为快速上手教程，测试音频（[中文](https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/vad_example.wav)，[英文](https://isv-data.oss-cn-hangzhou.aliyuncs.com/ics/MaaS/ASR/test_audio/asr_example_en.wav)）
+
+### 可执行命令行
+在命令行终端执行：
+
+```shell
+funasr +model=paraformer-zh +vad_model="fsmn-vad" +punc_model="ct-punc" +input=vad_example.wav
+```
+
+注：支持单条音频文件识别，也支持文件列表，列表为kaldi风格wav.scp：`wav_id   wav_path`
+
+### python示例
+#### 非实时语音识别
+```python
+from funasr import AutoModel
+# paraformer-zh is a multi-functional asr model
+# use vad, punc, spk or not as you need
+model = AutoModel(model="paraformer-zh", model_revision="v2.0.4",
+                  vad_model="fsmn-vad", vad_model_revision="v2.0.4",
+                  punc_model="ct-punc-c", punc_model_revision="v2.0.4",
+                  # spk_model="cam++", spk_model_revision="v2.0.2",
+                  )
+res = model.generate(input=f"{model.model_path}/example/asr_example.wav", 
+            batch_size_s=300, 
+            hotword='魔搭')
+print(res)
+```
+注：`model_hub`：表示模型仓库，`ms`为选择modelscope下载，`hf`为选择huggingface下载。
+
+#### 实时语音识别
+
+```python
+from funasr import AutoModel
+
+chunk_size = [0, 10, 5] #[0, 10, 5] 600ms, [0, 8, 4] 480ms
+encoder_chunk_look_back = 4 #number of chunks to lookback for encoder self-attention
+decoder_chunk_look_back = 1 #number of encoder chunks to lookback for decoder cross-attention
+
+model = AutoModel(model="paraformer-zh-streaming", model_revision="v2.0.4")
+
+import soundfile
+import os
+
+wav_file = os.path.join(model.model_path, "example/asr_example.wav")
+speech, sample_rate = soundfile.read(wav_file)
+chunk_stride = chunk_size[1] * 960 # 600ms
+
+cache = {}
+total_chunk_num = int(len((speech)-1)/chunk_stride+1)
+for i in range(total_chunk_num):
+    speech_chunk = speech[i*chunk_stride:(i+1)*chunk_stride]
+    is_final = i == total_chunk_num - 1
+    res = model.generate(input=speech_chunk, cache=cache, is_final=is_final, chunk_size=chunk_size, encoder_chunk_look_back=encoder_chunk_look_back, decoder_chunk_look_back=decoder_chunk_look_back)
+    print(res)
+```
+
+注：`chunk_size`为流式延时配置，`[0,10,5]`表示上屏实时出字粒度为`10*60=600ms`，未来信息为`5*60=300ms`。每次推理输入为`600ms`（采样点数为`16000*0.6=960`），输出为对应文字，最后一个语音片段输入需要设置`is_final=True`来强制输出最后一个字。
+
+#### 语音端点检测（非实时）
+```python
+from funasr import AutoModel
+
+model = AutoModel(model="fsmn-vad", model_revision="v2.0.4")
+
+wav_file = f"{model.model_path}/example/asr_example.wav"
+res = model.generate(input=wav_file)
+print(res)
+```
+
+#### 语音端点检测（实时）
+```python
+from funasr import AutoModel
+
+chunk_size = 200 # ms
+model = AutoModel(model="fsmn-vad", model_revision="v2.0.4")
+
+import soundfile
+
+wav_file = f"{model.model_path}/example/vad_example.wav"
+speech, sample_rate = soundfile.read(wav_file)
+chunk_stride = int(chunk_size * sample_rate / 1000)
+
+cache = {}
+total_chunk_num = int(len((speech)-1)/chunk_stride+1)
+for i in range(total_chunk_num):
+    speech_chunk = speech[i*chunk_stride:(i+1)*chunk_stride]
+    is_final = i == total_chunk_num - 1
+    res = model.generate(input=speech_chunk, cache=cache, is_final=is_final, chunk_size=chunk_size)
+    if len(res[0]["value"]):
+        print(res)
+```
+
+#### 标点恢复
+```python
+from funasr import AutoModel
+
+model = AutoModel(model="ct-punc", model_revision="v2.0.4")
+
+res = model.generate(input="那今天的会就到这里吧 happy new year 明年见")
+print(res)
+```
+
+#### 时间戳预测
+```python
+from funasr import AutoModel
+
+model = AutoModel(model="fa-zh", model_revision="v2.0.4")
+
+wav_file = f"{model.model_path}/example/asr_example.wav"
+text_file = f"{model.model_path}/example/text.txt"
+res = model.generate(input=(wav_file, text_file), data_type=("sound", "text"))
+print(res)
+```
+
+更多详细用法（[示例](https://github.com/alibaba-damo-academy/FunASR/tree/main/examples/industrial_data_pretraining)）
+
+
+## 微调
+
+详细用法（[示例](https://github.com/alibaba-damo-academy/FunASR/tree/main/examples/industrial_data_pretraining)）
+
+
+## 相关论文以及引用信息
+
+```BibTeX
+@article{shi2023seaco,
+  title={SeACo-Paraformer: A Non-Autoregressive ASR System with Flexible and Effective Hotword Customization Ability},
+  author={Shi, Xian and Yang, Yexin and Li, Zerui and Zhang, Shiliang},
+  journal={arXiv preprint arXiv:2308.03266 (accepted by ICASSP2024)},
+  year={2023}
+}
+```

moyoyo_asr_models/speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/am.mvn

@@ -0,0 +1,8 @@
+<Nnet> 
+<Splice> 560 560
+[ 0 ]
+<AddShift> 560 560 
+<LearnRateCoef> 0 [ -8.311879 -8.600912 -9.615928 -10.43595 -11.21292 -11.88333 -12.36243 -12.63706 -12.8818 -12.83066 -12.89103 -12.95666 -13.19763 -13.40598 -13.49113 -13.5546 -13.55639 -13.51915 -13.68284 -13.53289 -13.42107 -13.65519 -13.50713 -13.75251 -13.76715 -13.87408 -13.73109 -13.70412 -13.56073 -13.53488 -13.54895 -13.56228 -13.59408 -13.62047 -13.64198 -13.66109 -13.62669 -13.58297 -13.57387 -13.4739 -13.53063 -13.48348 -13.61047 -13.64716 -13.71546 -13.79184 -13.90614 -14.03098 -14.18205 -14.35881 -14.48419 -14.60172 -14.70591 -14.83362 -14.92122 -15.00622 -15.05122 -15.03119 -14.99028 -14.92302 -14.86927 -14.82691 -14.7972 -14.76909 -14.71356 -14.61277 -14.51696 -14.42252 -14.36405 -14.30451 -14.23161 -14.19851 -14.16633 -14.15649 -14.10504 -13.99518 -13.79562 -13.3996 -12.7767 -11.71208 -8.311879 -8.600912 -9.615928 -10.43595 -11.21292 -11.88333 -12.36243 -12.63706 -12.8818 -12.83066 -12.89103 -12.95666 -13.19763 -13.40598 -13.49113 -13.5546 -13.55639 -13.51915 -13.68284 -13.53289 -13.42107 -13.65519 -13.50713 -13.75251 -13.76715 -13.87408 -13.73109 -13.70412 -13.56073 -13.53488 -13.54895 -13.56228 -13.59408 -13.62047 -13.64198 -13.66109 -13.62669 -13.58297 -13.57387 -13.4739 -13.53063 -13.48348 -13.61047 -13.64716 -13.71546 -13.79184 -13.90614 -14.03098 -14.18205 -14.35881 -14.48419 -14.60172 -14.70591 -14.83362 -14.92122 -15.00622 -15.05122 -15.03119 -14.99028 -14.92302 -14.86927 -14.82691 -14.7972 -14.76909 -14.71356 -14.61277 -14.51696 -14.42252 -14.36405 -14.30451 -14.23161 -14.19851 -14.16633 -14.15649 -14.10504 -13.99518 -13.79562 -13.3996 -12.7767 -11.71208 -8.311879 -8.600912 -9.615928 -10.43595 -11.21292 -11.88333 -12.36243 -12.63706 -12.8818 -12.83066 -12.89103 -12.95666 -13.19763 -13.40598 -13.49113 -13.5546 -13.55639 -13.51915 -13.68284 -13.53289 -13.42107 -13.65519 -13.50713 -13.75251 -13.76715 -13.87408 -13.73109 -13.70412 -13.56073 -13.53488 -13.54895 -13.56228 -13.59408 -13.62047 -13.64198 -13.66109 -13.62669 -13.58297 -13.57387 -13.4739 -13.53063 -13.48348 -13.61047 -13.64716 -13.71546 -13.79184 -13.90614 -14.03098 -14.18205 -14.35881 -14.48419 -14.60172 -14.70591 -14.83362 -14.92122 -15.00622 -15.05122 -15.03119 -14.99028 -14.92302 -14.86927 -14.82691 -14.7972 -14.76909 -14.71356 -14.61277 -14.51696 -14.42252 -14.36405 -14.30451 -14.23161 -14.19851 -14.16633 -14.15649 -14.10504 -13.99518 -13.79562 -13.3996 -12.7767 -11.71208 -8.311879 -8.600912 -9.615928 -10.43595 -11.21292 -11.88333 -12.36243 -12.63706 -12.8818 -12.83066 -12.89103 -12.95666 -13.19763 -13.40598 -13.49113 -13.5546 -13.55639 -13.51915 -13.68284 -13.53289 -13.42107 -13.65519 -13.50713 -13.75251 -13.76715 -13.87408 -13.73109 -13.70412 -13.56073 -13.53488 -13.54895 -13.56228 -13.59408 -13.62047 -13.64198 -13.66109 -13.62669 -13.58297 -13.57387 -13.4739 -13.53063 -13.48348 -13.61047 -13.64716 -13.71546 -13.79184 -13.90614 -14.03098 -14.18205 -14.35881 -14.48419 -14.60172 -14.70591 -14.83362 -14.92122 -15.00622 -15.05122 -15.03119 -14.99028 -14.92302 -14.86927 -14.82691 -14.7972 -14.76909 -14.71356 -14.61277 -14.51696 -14.42252 -14.36405 -14.30451 -14.23161 -14.19851 -14.16633 -14.15649 -14.10504 -13.99518 -13.79562 -13.3996 -12.7767 -11.71208 -8.311879 -8.600912 -9.615928 -10.43595 -11.21292 -11.88333 -12.36243 -12.63706 -12.8818 -12.83066 -12.89103 -12.95666 -13.19763 -13.40598 -13.49113 -13.5546 -13.55639 -13.51915 -13.68284 -13.53289 -13.42107 -13.65519 -13.50713 -13.75251 -13.76715 -13.87408 -13.73109 -13.70412 -13.56073 -13.53488 -13.54895 -13.56228 -13.59408 -13.62047 -13.64198 -13.66109 -13.62669 -13.58297 -13.57387 -13.4739 -13.53063 -13.48348 -13.61047 -13.64716 -13.71546 -13.79184 -13.90614 -14.03098 -14.18205 -14.35881 -14.48419 -14.60172 -14.70591 -14.83362 -14.92122 -15.00622 -15.05122 -15.03119 -14.99028 -14.92302 -14.86927 -14.82691 -14.7972 -14.76909 -14.71356 -14.61277 -14.51696 -14.42252 -14.36405 -14.30451 -14.23161 -14.19851 -14.16633 -14.15649 -14.10504 -13.99518 -13.79562 -13.3996 -12.7767 -11.71208 -8.311879 -8.600912 -9.615928 -10.43595 -11.21292 -11.88333 -12.36243 -12.63706 -12.8818 -12.83066 -12.89103 -12.95666 -13.19763 -13.40598 -13.49113 -13.5546 -13.55639 -13.51915 -13.68284 -13.53289 -13.42107 -13.65519 -13.50713 -13.75251 -13.76715 -13.87408 -13.73109 -13.70412 -13.56073 -13.53488 -13.54895 -13.56228 -13.59408 -13.62047 -13.64198 -13.66109 -13.62669 -13.58297 -13.57387 -13.4739 -13.53063 -13.48348 -13.61047 -13.64716 -13.71546 -13.79184 -13.90614 -14.03098 -14.18205 -14.35881 -14.48419 -14.60172 -14.70591 -14.83362 -14.92122 -15.00622 -15.05122 -15.03119 -14.99028 -14.92302 -14.86927 -14.82691 -14.7972 -14.76909 -14.71356 -14.61277 -14.51696 -14.42252 -14.36405 -14.30451 -14.23161 -14.19851 -14.16633 -14.15649 -14.10504 -13.99518 -13.79562 -13.3996 -12.7767 -11.71208 -8.311879 -8.600912 -9.615928 -10.43595 -11.21292 -11.88333 -12.36243 -12.63706 -12.8818 -12.83066 -12.89103 -12.95666 -13.19763 -13.40598 -13.49113 -13.5546 -13.55639 -13.51915 -13.68284 -13.53289 -13.42107 -13.65519 -13.50713 -13.75251 -13.76715 -13.87408 -13.73109 -13.70412 -13.56073 -13.53488 -13.54895 -13.56228 -13.59408 -13.62047 -13.64198 -13.66109 -13.62669 -13.58297 -13.57387 -13.4739 -13.53063 -13.48348 -13.61047 -13.64716 -13.71546 -13.79184 -13.90614 -14.03098 -14.18205 -14.35881 -14.48419 -14.60172 -14.70591 -14.83362 -14.92122 -15.00622 -15.05122 -15.03119 -14.99028 -14.92302 -14.86927 -14.82691 -14.7972 -14.76909 -14.71356 -14.61277 -14.51696 -14.42252 -14.36405 -14.30451 -14.23161 -14.19851 -14.16633 -14.15649 -14.10504 -13.99518 -13.79562 -13.3996 -12.7767 -11.71208 ]
+<Rescale> 560 560
+<LearnRateCoef> 0 [ 0.155775 0.154484 0.1527379 0.1518718 0.1506028 0.1489256 0.147067 0.1447061 0.1436307 0.1443568 0.1451849 0.1455157 0.1452821 0.1445717 0.1439195 0.1435867 0.1436018 0.1438781 0.1442086 0.1448844 0.1454756 0.145663 0.146268 0.1467386 0.1472724 0.147664 0.1480913 0.1483739 0.1488841 0.1493636 0.1497088 0.1500379 0.1502916 0.1505389 0.1506787 0.1507102 0.1505992 0.1505445 0.1505938 0.1508133 0.1509569 0.1512396 0.1514625 0.1516195 0.1516156 0.1515561 0.1514966 0.1513976 0.1512612 0.151076 0.1510596 0.1510431 0.151077 0.1511168 0.1511917 0.151023 0.1508045 0.1505885 0.1503493 0.1502373 0.1501726 0.1500762 0.1500065 0.1499782 0.150057 0.1502658 0.150469 0.1505335 0.1505505 0.1505328 0.1504275 0.1502438 0.1499674 0.1497118 0.1494661 0.1493102 0.1493681 0.1495501 0.1499738 0.1509654 0.155775 0.154484 0.1527379 0.1518718 0.1506028 0.1489256 0.147067 0.1447061 0.1436307 0.1443568 0.1451849 0.1455157 0.1452821 0.1445717 0.1439195 0.1435867 0.1436018 0.1438781 0.1442086 0.1448844 0.1454756 0.145663 0.146268 0.1467386 0.1472724 0.147664 0.1480913 0.1483739 0.1488841 0.1493636 0.1497088 0.1500379 0.1502916 0.1505389 0.1506787 0.1507102 0.1505992 0.1505445 0.1505938 0.1508133 0.1509569 0.1512396 0.1514625 0.1516195 0.1516156 0.1515561 0.1514966 0.1513976 0.1512612 0.151076 0.1510596 0.1510431 0.151077 0.1511168 0.1511917 0.151023 0.1508045 0.1505885 0.1503493 0.1502373 0.1501726 0.1500762 0.1500065 0.1499782 0.150057 0.1502658 0.150469 0.1505335 0.1505505 0.1505328 0.1504275 0.1502438 0.1499674 0.1497118 0.1494661 0.1493102 0.1493681 0.1495501 0.1499738 0.1509654 0.155775 0.154484 0.1527379 0.1518718 0.1506028 0.1489256 0.147067 0.1447061 0.1436307 0.1443568 0.1451849 0.1455157 0.1452821 0.1445717 0.1439195 0.1435867 0.1436018 0.1438781 0.1442086 0.1448844 0.1454756 0.145663 0.146268 0.1467386 0.1472724 0.147664 0.1480913 0.1483739 0.1488841 0.1493636 0.1497088 0.1500379 0.1502916 0.1505389 0.1506787 0.1507102 0.1505992 0.1505445 0.1505938 0.1508133 0.1509569 0.1512396 0.1514625 0.1516195 0.1516156 0.1515561 0.1514966 0.1513976 0.1512612 0.151076 0.1510596 0.1510431 0.151077 0.1511168 0.1511917 0.151023 0.1508045 0.1505885 0.1503493 0.1502373 0.1501726 0.1500762 0.1500065 0.1499782 0.150057 0.1502658 0.150469 0.1505335 0.1505505 0.1505328 0.1504275 0.1502438 0.1499674 0.1497118 0.1494661 0.1493102 0.1493681 0.1495501 0.1499738 0.1509654 0.155775 0.154484 0.1527379 0.1518718 0.1506028 0.1489256 0.147067 0.1447061 0.1436307 0.1443568 0.1451849 0.1455157 0.1452821 0.1445717 0.1439195 0.1435867 0.1436018 0.1438781 0.1442086 0.1448844 0.1454756 0.145663 0.146268 0.1467386 0.1472724 0.147664 0.1480913 0.1483739 0.1488841 0.1493636 0.1497088 0.1500379 0.1502916 0.1505389 0.1506787 0.1507102 0.1505992 0.1505445 0.1505938 0.1508133 0.1509569 0.1512396 0.1514625 0.1516195 0.1516156 0.1515561 0.1514966 0.1513976 0.1512612 0.151076 0.1510596 0.1510431 0.151077 0.1511168 0.1511917 0.151023 0.1508045 0.1505885 0.1503493 0.1502373 0.1501726 0.1500762 0.1500065 0.1499782 0.150057 0.1502658 0.150469 0.1505335 0.1505505 0.1505328 0.1504275 0.1502438 0.1499674 0.1497118 0.1494661 0.1493102 0.1493681 0.1495501 0.1499738 0.1509654 0.155775 0.154484 0.1527379 0.1518718 0.1506028 0.1489256 0.147067 0.1447061 0.1436307 0.1443568 0.1451849 0.1455157 0.1452821 0.1445717 0.1439195 0.1435867 0.1436018 0.1438781 0.1442086 0.1448844 0.1454756 0.145663 0.146268 0.1467386 0.1472724 0.147664 0.1480913 0.1483739 0.1488841 0.1493636 0.1497088 0.1500379 0.1502916 0.1505389 0.1506787 0.1507102 0.1505992 0.1505445 0.1505938 0.1508133 0.1509569 0.1512396 0.1514625 0.1516195 0.1516156 0.1515561 0.1514966 0.1513976 0.1512612 0.151076 0.1510596 0.1510431 0.151077 0.1511168 0.1511917 0.151023 0.1508045 0.1505885 0.1503493 0.1502373 0.1501726 0.1500762 0.1500065 0.1499782 0.150057 0.1502658 0.150469 0.1505335 0.1505505 0.1505328 0.1504275 0.1502438 0.1499674 0.1497118 0.1494661 0.1493102 0.1493681 0.1495501 0.1499738 0.1509654 0.155775 0.154484 0.1527379 0.1518718 0.1506028 0.1489256 0.147067 0.1447061 0.1436307 0.1443568 0.1451849 0.1455157 0.1452821 0.1445717 0.1439195 0.1435867 0.1436018 0.1438781 0.1442086 0.1448844 0.1454756 0.145663 0.146268 0.1467386 0.1472724 0.147664 0.1480913 0.1483739 0.1488841 0.1493636 0.1497088 0.1500379 0.1502916 0.1505389 0.1506787 0.1507102 0.1505992 0.1505445 0.1505938 0.1508133 0.1509569 0.1512396 0.1514625 0.1516195 0.1516156 0.1515561 0.1514966 0.1513976 0.1512612 0.151076 0.1510596 0.1510431 0.151077 0.1511168 0.1511917 0.151023 0.1508045 0.1505885 0.1503493 0.1502373 0.1501726 0.1500762 0.1500065 0.1499782 0.150057 0.1502658 0.150469 0.1505335 0.1505505 0.1505328 0.1504275 0.1502438 0.1499674 0.1497118 0.1494661 0.1493102 0.1493681 0.1495501 0.1499738 0.1509654 0.155775 0.154484 0.1527379 0.1518718 0.1506028 0.1489256 0.147067 0.1447061 0.1436307 0.1443568 0.1451849 0.1455157 0.1452821 0.1445717 0.1439195 0.1435867 0.1436018 0.1438781 0.1442086 0.1448844 0.1454756 0.145663 0.146268 0.1467386 0.1472724 0.147664 0.1480913 0.1483739 0.1488841 0.1493636 0.1497088 0.1500379 0.1502916 0.1505389 0.1506787 0.1507102 0.1505992 0.1505445 0.1505938 0.1508133 0.1509569 0.1512396 0.1514625 0.1516195 0.1516156 0.1515561 0.1514966 0.1513976 0.1512612 0.151076 0.1510596 0.1510431 0.151077 0.1511168 0.1511917 0.151023 0.1508045 0.1505885 0.1503493 0.1502373 0.1501726 0.1500762 0.1500065 0.1499782 0.150057 0.1502658 0.150469 0.1505335 0.1505505 0.1505328 0.1504275 0.1502438 0.1499674 0.1497118 0.1494661 0.1493102 0.1493681 0.1495501 0.1499738 0.1509654 ]
+</Nnet> 

moyoyo_asr_models/speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/asr_example_hotword.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:51792bc95be33075c1a8abb9afb76ad9f72943e84cd723cc8825b2678799b004
+size 253642

moyoyo_asr_models/speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/config.yaml

@@ -0,0 +1,160 @@
+# This is an example that demonstrates how to configure a model file.
+# You can modify the configuration according to your own requirements.
+
+# to print the register_table:
+# from funasr.utils.register import registry_tables
+# registry_tables.print()
+
+# network architecture
+model: SeacoParaformer
+model_conf:
+    ctc_weight: 0.0
+    lsm_weight: 0.1
+    length_normalized_loss: true
+    predictor_weight: 1.0
+    predictor_bias: 1
+    sampling_ratio: 0.75
+    inner_dim: 512
+    bias_encoder_type: lstm
+    bias_encoder_bid: false
+    seaco_lsm_weight: 0.1
+    seaco_length_normal: true
+    train_decoder: true
+    NO_BIAS: 8377
+
+# encoder
+encoder: SANMEncoder
+encoder_conf:
+    output_size: 512
+    attention_heads: 4
+    linear_units: 2048
+    num_blocks: 50
+    dropout_rate: 0.1
+    positional_dropout_rate: 0.1
+    attention_dropout_rate: 0.1
+    input_layer: pe
+    pos_enc_class: SinusoidalPositionEncoder
+    normalize_before: true
+    kernel_size: 11
+    sanm_shfit: 0
+    selfattention_layer_type: sanm
+
+# decoder
+decoder: ParaformerSANMDecoder
+decoder_conf:
+    attention_heads: 4
+    linear_units: 2048
+    num_blocks: 16
+    dropout_rate: 0.1
+    positional_dropout_rate: 0.1
+    self_attention_dropout_rate: 0.1
+    src_attention_dropout_rate: 0.1
+    att_layer_num: 16
+    kernel_size: 11
+    sanm_shfit: 0
+
+# seaco decoder
+seaco_decoder: ParaformerSANMDecoder
+seaco_decoder_conf:
+    attention_heads: 4
+    linear_units: 1024
+    num_blocks: 4
+    dropout_rate: 0.1
+    positional_dropout_rate: 0.1
+    self_attention_dropout_rate: 0.1
+    src_attention_dropout_rate: 0.1
+    kernel_size: 21
+    sanm_shfit: 0
+    use_output_layer: false
+    wo_input_layer: true
+
+predictor: CifPredictorV3
+predictor_conf:
+    idim: 512
+    threshold: 1.0
+    l_order: 1
+    r_order: 1
+    tail_threshold: 0.45
+    smooth_factor2: 0.25
+    noise_threshold2: 0.01
+    upsample_times: 3
+    use_cif1_cnn: false
+    upsample_type: cnn_blstm
+
+# frontend related
+frontend: WavFrontend
+frontend_conf:
+    fs: 16000
+    window: hamming
+    n_mels: 80
+    frame_length: 25
+    frame_shift: 10
+    lfr_m: 7
+    lfr_n: 6
+    dither: 0.0
+
+specaug: SpecAugLFR
+specaug_conf:
+    apply_time_warp: false
+    time_warp_window: 5
+    time_warp_mode: bicubic
+    apply_freq_mask: true
+    freq_mask_width_range:
+    - 0
+    - 30
+    lfr_rate: 6
+    num_freq_mask: 1
+    apply_time_mask: true
+    time_mask_width_range:
+    - 0
+    - 12
+    num_time_mask: 1
+
+train_conf:
+  accum_grad: 1
+  grad_clip: 5
+  max_epoch: 150
+  val_scheduler_criterion:
+      - valid
+      - acc
+  best_model_criterion:
+  -   - valid
+      - acc
+      - max
+  keep_nbest_models: 10
+  log_interval: 50
+  unused_parameters: true
+
+optim: adam
+optim_conf:
+   lr: 0.0005
+scheduler: warmuplr
+scheduler_conf:
+   warmup_steps: 30000
+
+dataset: AudioDatasetHotword
+dataset_conf:
+    seaco_id: 8377
+    index_ds: IndexDSJsonl
+    batch_sampler: DynamicBatchLocalShuffleSampler
+    batch_type: example # example or length
+    batch_size: 1 # if batch_type is example, batch_size is the numbers of samples; if length, batch_size is source_token_len+target_token_len;
+    max_token_length: 2048 # filter samples if source_token_len+target_token_len > max_token_length,
+    buffer_size: 500
+    shuffle: True
+    num_workers: 0
+
+tokenizer: CharTokenizer
+tokenizer_conf:
+  unk_symbol: <unk>
+  split_with_space: true
+
+
+ctc_conf:
+    dropout_rate: 0.0
+    ctc_type: builtin
+    reduce: true
+    ignore_nan_grad: true
+    
+normalize: null
+

moyoyo_asr_models/speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/configuration.json

@@ -0,0 +1,14 @@
+{
+  "framework": "pytorch",
+  "task" : "auto-speech-recognition",
+  "model": {"type" : "funasr"},
+  "pipeline": {"type":"funasr-pipeline"},
+  "model_name_in_hub": {
+    "ms":"iic/speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch", 
+    "hf":""},
+  "file_path_metas": {
+    "init_param":"model.pt", 
+    "config":"config.yaml",
+    "tokenizer_conf": {"token_list": "tokens.json", "seg_dict_file": "seg_dict"},
+    "frontend_conf":{"cmvn_file": "am.mvn"}}
+}

moyoyo_asr_models/speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/example/asr_example.wav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2ffa478de2cd570dd54e8762008cd6bbde9871fd79757f1cdbbec7d6b7b49274
+size 144770

moyoyo_asr_models/speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/example/hotword.txt

@@ -0,0 +1 @@
+魔搭

moyoyo_asr_models/speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3d491689244ec5dfbf9170ef3827c358aa10f1f20e42a7c59e15e688647946d1
+size 989763045

tests/test_prompt.ipynb

@@ -0,0 +1,142 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import sys\n",
+    "sys.path.append(\"..\")\n",
+    "import config"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "PROMPT = \"\"\"你是一位精通简体中文的专业翻译，尤其擅长将专业学术论文翻译成浅显易懂的科普文章。请你帮我将以下英文段落翻译成中文，风格与中文科普读物相似。\n",
+    "\n",
+    "规则：\n",
+    "- 翻译时要准确传达原文的事实和背景。\n",
+    "- 即使上意译也要保留原始段落格式，以及保留术语，例如 FLAC，JPEG 等。保留公司缩写，例如 Microsoft, Amazon, OpenAI 等。\n",
+    "- 人名不翻译\n",
+    "- 全角括号换成半角括号，并在左括号前面加半角空格，右括号后面加半角空格。\n",
+    "- 在翻译专业术语时，第一次出现时要在括号里面写上英文原文，例如：“生成式 AI (Generative AI)”，之后就可以只写中文了。\n",
+    "- 以下是常见的 AI 相关术语词汇对应表（English -> 中文）：\n",
+    "  * Transformer -> Transformer\n",
+    "  * Token -> Token\n",
+    "  * LLM/Large Language Model -> 大语言模型\n",
+    "  * Zero-shot -> 零样本\n",
+    "  * Few-shot -> 少样本\n",
+    "  * AI Agent -> AI 智能体\n",
+    "  * AGI -> 通用人工智能\"\"\""
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from transcribe.helpers.translator import QwenTranslator"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "llama_context: n_ctx_per_seq (512) < n_ctx_train (131072) -- the full capacity of the model will not be utilized\n",
+      "ggml_metal_init: skipping kernel_get_rows_bf16                     (not supported)\n",
+      "ggml_metal_init: skipping kernel_mul_mv_bf16_f32                   (not supported)\n",
+      "ggml_metal_init: skipping kernel_mul_mv_bf16_f32_1row              (not supported)\n",
+      "ggml_metal_init: skipping kernel_mul_mv_bf16_f32_l4                (not supported)\n",
+      "ggml_metal_init: skipping kernel_mul_mv_bf16_bf16                  (not supported)\n",
+      "ggml_metal_init: skipping kernel_mul_mv_id_bf16_f32                (not supported)\n",
+      "ggml_metal_init: skipping kernel_mul_mm_bf16_f32                   (not supported)\n",
+      "ggml_metal_init: skipping kernel_mul_mm_id_bf16_f32                (not supported)\n",
+      "ggml_metal_init: skipping kernel_flash_attn_ext_bf16_h64           (not supported)\n",
+      "ggml_metal_init: skipping kernel_flash_attn_ext_bf16_h80           (not supported)\n",
+      "ggml_metal_init: skipping kernel_flash_attn_ext_bf16_h96           (not supported)\n",
+      "ggml_metal_init: skipping kernel_flash_attn_ext_bf16_h112          (not supported)\n",
+      "ggml_metal_init: skipping kernel_flash_attn_ext_bf16_h128          (not supported)\n",
+      "ggml_metal_init: skipping kernel_flash_attn_ext_bf16_h192          (not supported)\n",
+      "ggml_metal_init: skipping kernel_flash_attn_ext_bf16_hk192_hv128   (not supported)\n",
+      "ggml_metal_init: skipping kernel_flash_attn_ext_bf16_h256          (not supported)\n",
+      "ggml_metal_init: skipping kernel_flash_attn_ext_vec_bf16_h96       (not supported)\n",
+      "ggml_metal_init: skipping kernel_flash_attn_ext_vec_bf16_h128      (not supported)\n",
+      "ggml_metal_init: skipping kernel_flash_attn_ext_vec_bf16_h192      (not supported)\n",
+      "ggml_metal_init: skipping kernel_flash_attn_ext_vec_bf16_hk192_hv128 (not supported)\n",
+      "ggml_metal_init: skipping kernel_flash_attn_ext_vec_bf16_h256      (not supported)\n",
+      "ggml_metal_init: skipping kernel_cpy_f32_bf16                      (not supported)\n",
+      "ggml_metal_init: skipping kernel_cpy_bf16_f32                      (not supported)\n",
+      "ggml_metal_init: skipping kernel_cpy_bf16_bf16                     (not supported)\n"
+     ]
+    }
+   ],
+   "source": [
+    "translator = QwenTranslator(config.LLM_LARGE_MODEL_PATH, PROMPT, PROMPT)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "metadata": {},
+   "outputs": [
+    {
+     "ename": "TypeError",
+     "evalue": "'<=' not supported between instances of 'str' and 'int'",
+     "output_type": "error",
+     "traceback": [
+      "\u001b[31m---------------------------------------------------------------------------\u001b[39m",
+      "\u001b[31mTypeError\u001b[39m                                 Traceback (most recent call last)",
+      "\u001b[36mCell\u001b[39m\u001b[36m \u001b[39m\u001b[32mIn[7]\u001b[39m\u001b[32m, line 1\u001b[39m\n\u001b[32m----> \u001b[39m\u001b[32m1\u001b[39m \u001b[43mtranslator\u001b[49m\u001b[43m(\u001b[49m\u001b[33;43m\"\u001b[39;49m\u001b[33;43m欢迎来到go sim大会\u001b[39;49m\u001b[33;43m\"\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[33;43m\"\u001b[39;49m\u001b[33;43mzh\u001b[39;49m\u001b[33;43m\"\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[33;43m'\u001b[39;49m\u001b[33;43men\u001b[39;49m\u001b[33;43m'\u001b[39;49m\u001b[43m)\u001b[49m\n",
+      "\u001b[36mFile \u001b[39m\u001b[32m~/Codes/trans/tests/../transcribe/helpers/translator.py:30\u001b[39m, in \u001b[36mQwenTranslator.__call__\u001b[39m\u001b[34m(self, prompt, *args, **kwargs)\u001b[39m\n\u001b[32m     29\u001b[39m \u001b[38;5;28;01mdef\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[34m__call__\u001b[39m(\u001b[38;5;28mself\u001b[39m, prompt,*args, **kwargs):\n\u001b[32m---> \u001b[39m\u001b[32m30\u001b[39m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[43m.\u001b[49m\u001b[43mllm\u001b[49m\u001b[43m(\u001b[49m\n\u001b[32m     31\u001b[39m \u001b[43m        \u001b[49m\u001b[43mprompt\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m     32\u001b[39m \u001b[43m        \u001b[49m\u001b[43m*\u001b[49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m     33\u001b[39m \u001b[43m        \u001b[49m\u001b[43m*\u001b[49m\u001b[43m*\u001b[49m\u001b[43mkwargs\u001b[49m\n\u001b[32m     34\u001b[39m \u001b[43m    \u001b[49m\u001b[43m)\u001b[49m\n",
+      "\u001b[36mFile \u001b[39m\u001b[32m~/Codes/trans/.venv/lib/python3.11/site-packages/llama_cpp/llama.py:1902\u001b[39m, in \u001b[36mLlama.__call__\u001b[39m\u001b[34m(self, prompt, suffix, max_tokens, temperature, top_p, min_p, typical_p, logprobs, echo, stop, frequency_penalty, presence_penalty, repeat_penalty, top_k, stream, seed, tfs_z, mirostat_mode, mirostat_tau, mirostat_eta, model, stopping_criteria, logits_processor, grammar, logit_bias)\u001b[39m\n\u001b[32m   1838\u001b[39m \u001b[38;5;28;01mdef\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[34m__call__\u001b[39m(\n\u001b[32m   1839\u001b[39m     \u001b[38;5;28mself\u001b[39m,\n\u001b[32m   1840\u001b[39m     prompt: \u001b[38;5;28mstr\u001b[39m,\n\u001b[32m   (...)\u001b[39m\u001b[32m   1864\u001b[39m     logit_bias: Optional[Dict[\u001b[38;5;28mint\u001b[39m, \u001b[38;5;28mfloat\u001b[39m]] = \u001b[38;5;28;01mNone\u001b[39;00m,\n\u001b[32m   1865\u001b[39m ) -> Union[CreateCompletionResponse, Iterator[CreateCompletionStreamResponse]]:\n\u001b[32m   1866\u001b[39m \u001b[38;5;250m    \u001b[39m\u001b[33;03m\"\"\"Generate text from a prompt.\u001b[39;00m\n\u001b[32m   1867\u001b[39m \n\u001b[32m   1868\u001b[39m \u001b[33;03m    Args:\u001b[39;00m\n\u001b[32m   (...)\u001b[39m\u001b[32m   1900\u001b[39m \u001b[33;03m        Response object containing the generated text.\u001b[39;00m\n\u001b[32m   1901\u001b[39m \u001b[33;03m    \"\"\"\u001b[39;00m\n\u001b[32m-> \u001b[39m\u001b[32m1902\u001b[39m     \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[43m.\u001b[49m\u001b[43mcreate_completion\u001b[49m\u001b[43m(\u001b[49m\n\u001b[32m   1903\u001b[39m \u001b[43m        \u001b[49m\u001b[43mprompt\u001b[49m\u001b[43m=\u001b[49m\u001b[43mprompt\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1904\u001b[39m \u001b[43m        \u001b[49m\u001b[43msuffix\u001b[49m\u001b[43m=\u001b[49m\u001b[43msuffix\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1905\u001b[39m \u001b[43m        \u001b[49m\u001b[43mmax_tokens\u001b[49m\u001b[43m=\u001b[49m\u001b[43mmax_tokens\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1906\u001b[39m \u001b[43m        \u001b[49m\u001b[43mtemperature\u001b[49m\u001b[43m=\u001b[49m\u001b[43mtemperature\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1907\u001b[39m \u001b[43m        \u001b[49m\u001b[43mtop_p\u001b[49m\u001b[43m=\u001b[49m\u001b[43mtop_p\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1908\u001b[39m \u001b[43m        \u001b[49m\u001b[43mmin_p\u001b[49m\u001b[43m=\u001b[49m\u001b[43mmin_p\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1909\u001b[39m \u001b[43m        \u001b[49m\u001b[43mtypical_p\u001b[49m\u001b[43m=\u001b[49m\u001b[43mtypical_p\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1910\u001b[39m \u001b[43m        \u001b[49m\u001b[43mlogprobs\u001b[49m\u001b[43m=\u001b[49m\u001b[43mlogprobs\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1911\u001b[39m \u001b[43m        \u001b[49m\u001b[43mecho\u001b[49m\u001b[43m=\u001b[49m\u001b[43mecho\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1912\u001b[39m \u001b[43m        \u001b[49m\u001b[43mstop\u001b[49m\u001b[43m=\u001b[49m\u001b[43mstop\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1913\u001b[39m \u001b[43m        \u001b[49m\u001b[43mfrequency_penalty\u001b[49m\u001b[43m=\u001b[49m\u001b[43mfrequency_penalty\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1914\u001b[39m \u001b[43m        \u001b[49m\u001b[43mpresence_penalty\u001b[49m\u001b[43m=\u001b[49m\u001b[43mpresence_penalty\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1915\u001b[39m \u001b[43m        \u001b[49m\u001b[43mrepeat_penalty\u001b[49m\u001b[43m=\u001b[49m\u001b[43mrepeat_penalty\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1916\u001b[39m \u001b[43m        \u001b[49m\u001b[43mtop_k\u001b[49m\u001b[43m=\u001b[49m\u001b[43mtop_k\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1917\u001b[39m \u001b[43m        \u001b[49m\u001b[43mstream\u001b[49m\u001b[43m=\u001b[49m\u001b[43mstream\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1918\u001b[39m \u001b[43m        \u001b[49m\u001b[43mseed\u001b[49m\u001b[43m=\u001b[49m\u001b[43mseed\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1919\u001b[39m \u001b[43m        \u001b[49m\u001b[43mtfs_z\u001b[49m\u001b[43m=\u001b[49m\u001b[43mtfs_z\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1920\u001b[39m \u001b[43m        \u001b[49m\u001b[43mmirostat_mode\u001b[49m\u001b[43m=\u001b[49m\u001b[43mmirostat_mode\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1921\u001b[39m \u001b[43m        \u001b[49m\u001b[43mmirostat_tau\u001b[49m\u001b[43m=\u001b[49m\u001b[43mmirostat_tau\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1922\u001b[39m \u001b[43m        \u001b[49m\u001b[43mmirostat_eta\u001b[49m\u001b[43m=\u001b[49m\u001b[43mmirostat_eta\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1923\u001b[39m \u001b[43m        \u001b[49m\u001b[43mmodel\u001b[49m\u001b[43m=\u001b[49m\u001b[43mmodel\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1924\u001b[39m \u001b[43m        \u001b[49m\u001b[43mstopping_criteria\u001b[49m\u001b[43m=\u001b[49m\u001b[43mstopping_criteria\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1925\u001b[39m \u001b[43m        \u001b[49m\u001b[43mlogits_processor\u001b[49m\u001b[43m=\u001b[49m\u001b[43mlogits_processor\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1926\u001b[39m \u001b[43m        \u001b[49m\u001b[43mgrammar\u001b[49m\u001b[43m=\u001b[49m\u001b[43mgrammar\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1927\u001b[39m \u001b[43m        \u001b[49m\u001b[43mlogit_bias\u001b[49m\u001b[43m=\u001b[49m\u001b[43mlogit_bias\u001b[49m\u001b[43m,\u001b[49m\n\u001b[32m   1928\u001b[39m \u001b[43m    \u001b[49m\u001b[43m)\u001b[49m\n",
+      "\u001b[36mFile \u001b[39m\u001b[32m~/Codes/trans/.venv/lib/python3.11/site-packages/llama_cpp/llama.py:1835\u001b[39m, in \u001b[36mLlama.create_completion\u001b[39m\u001b[34m(self, prompt, suffix, max_tokens, temperature, top_p, min_p, typical_p, logprobs, echo, stop, frequency_penalty, presence_penalty, repeat_penalty, top_k, stream, seed, tfs_z, mirostat_mode, mirostat_tau, mirostat_eta, model, stopping_criteria, logits_processor, grammar, logit_bias)\u001b[39m\n\u001b[32m   1833\u001b[39m     chunks: Iterator[CreateCompletionStreamResponse] = completion_or_chunks\n\u001b[32m   1834\u001b[39m     \u001b[38;5;28;01mreturn\u001b[39;00m chunks\n\u001b[32m-> \u001b[39m\u001b[32m1835\u001b[39m completion: Completion = \u001b[38;5;28;43mnext\u001b[39;49m\u001b[43m(\u001b[49m\u001b[43mcompletion_or_chunks\u001b[49m\u001b[43m)\u001b[49m  \u001b[38;5;66;03m# type: ignore\u001b[39;00m\n\u001b[32m   1836\u001b[39m \u001b[38;5;28;01mreturn\u001b[39;00m completion\n",
+      "\u001b[36mFile \u001b[39m\u001b[32m~/Codes/trans/.venv/lib/python3.11/site-packages/llama_cpp/llama.py:1275\u001b[39m, in \u001b[36mLlama._create_completion\u001b[39m\u001b[34m(self, prompt, suffix, max_tokens, temperature, top_p, min_p, typical_p, logprobs, echo, stop, frequency_penalty, presence_penalty, repeat_penalty, top_k, stream, seed, tfs_z, mirostat_mode, mirostat_tau, mirostat_eta, model, stopping_criteria, logits_processor, grammar, logit_bias)\u001b[39m\n\u001b[32m   1270\u001b[39m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mlen\u001b[39m(prompt_tokens) >= \u001b[38;5;28mself\u001b[39m._n_ctx:\n\u001b[32m   1271\u001b[39m     \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mValueError\u001b[39;00m(\n\u001b[32m   1272\u001b[39m         \u001b[33mf\u001b[39m\u001b[33m\"\u001b[39m\u001b[33mRequested tokens (\u001b[39m\u001b[38;5;132;01m{\u001b[39;00m\u001b[38;5;28mlen\u001b[39m(prompt_tokens)\u001b[38;5;132;01m}\u001b[39;00m\u001b[33m) exceed context window of \u001b[39m\u001b[38;5;132;01m{\u001b[39;00mllama_cpp.llama_n_ctx(\u001b[38;5;28mself\u001b[39m.ctx)\u001b[38;5;132;01m}\u001b[39;00m\u001b[33m\"\u001b[39m\n\u001b[32m   1273\u001b[39m     )\n\u001b[32m-> \u001b[39m\u001b[32m1275\u001b[39m \u001b[38;5;28;01mif\u001b[39;00m max_tokens \u001b[38;5;129;01mis\u001b[39;00m \u001b[38;5;28;01mNone\u001b[39;00m \u001b[38;5;129;01mor\u001b[39;00m \u001b[43mmax_tokens\u001b[49m\u001b[43m \u001b[49m\u001b[43m<\u001b[49m\u001b[43m=\u001b[49m\u001b[43m \u001b[49m\u001b[32;43m0\u001b[39;49m:\n\u001b[32m   1276\u001b[39m     \u001b[38;5;66;03m# Unlimited, depending on n_ctx.\u001b[39;00m\n\u001b[32m   1277\u001b[39m     max_tokens = \u001b[38;5;28mself\u001b[39m._n_ctx - \u001b[38;5;28mlen\u001b[39m(prompt_tokens)\n\u001b[32m   1279\u001b[39m \u001b[38;5;66;03m# Truncate max_tokens if requested tokens would exceed the context window\u001b[39;00m\n",
+      "\u001b[31mTypeError\u001b[39m: '<=' not supported between instances of 'str' and 'int'"
+     ]
+    }
+   ],
+   "source": [
+    "translator(\"欢迎来到go sim大会\", \"zh\", 'en')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": ".venv",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.11.11"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}

transcribe/helpers/funasr.py

@@ -12,8 +12,16 @@ class FunASR:
     def __init__(self, source_lange: str = 'en', warmup=True) -> None:
         self.source_lange = source_lange
 
+        model_dir = config.MODEL_DIR
+        asr_model_path = model_dir / 'speech_seaco_paraformer_large_asr_nat-zh-cn-16k-common-vocab8404-pytorch'
+        vad_model_path = model_dir / 'speech_fsmn_vad_zh-cn-16k-common-pytorch'
+        punc_model_path = model_dir / 'punc_ct-transformer_cn-en-common-vocab471067-large'
         self.model = AutoModel(
-            model="paraformer-zh", vad_model="fsmn-vad", punc_model="ct-punc", log_level="ERROR",
+            model=asr_model_path.as_posix(),
+            vad_model=vad_model_path.as_posix(),
+            punc_model=punc_model_path.as_posix(),
+            log_level="ERROR",
+            disable_update=True
         )
         if warmup:
             self.warmup()

transcribe/pipelines/pipe_translate.py

@@ -2,7 +2,7 @@
 from .base import MetaItem, BasePipe, Segment
 from llama_cpp import Llama
 from ..helpers.translator import QwenTranslator
-from config import LLM_MODEL_PATH, LLM_SYS_PROMPT_EN, LLM_SYS_PROMPT_ZH, LLM_LARGE_MODEL_PATH, ALL_MARKERS
+from config import LLM_MODEL_PATH, LLM_SYS_PROMPT_EN, LLM_SYS_PROMPT_ZH, LLM_LARGE_MODEL_PATH, ALL_MARKERS, LLM_SYS_7B_PROMPT_EN, LLM_SYS_7B_PROMPT_ZH
 
 
 class TranslatePipe(BasePipe):
@@ -33,7 +33,7 @@ class Translate7BPipe(TranslatePipe):
     @classmethod
     def init(cls):
         if cls.translator is None:
-            cls.translator = QwenTranslator(LLM_LARGE_MODEL_PATH, LLM_SYS_PROMPT_EN, LLM_SYS_PROMPT_ZH)
+            cls.translator = QwenTranslator(LLM_LARGE_MODEL_PATH, LLM_SYS_7B_PROMPT_EN, LLM_SYS_7B_PROMPT_ZH)