Upload app.py with huggingface_hub

app.py

@@ -1,17 +1,15 @@
-import json
 import os
 import sys
-import threading
 import time
+import threading
+import warnings
 
 from huggingface_hub import snapshot_download
+import gradio as gr
 
-import warnings
 warnings.filterwarnings("ignore", category=FutureWarning)
 warnings.filterwarnings("ignore", category=UserWarning)
 
-import gradio as gr
-
 current_dir = os.path.dirname(os.path.abspath(__file__))
 sys.path.append(current_dir)
 sys.path.append(os.path.join(current_dir, "indextts"))
@@ -20,35 +18,36 @@ from indextts.infer import IndexTTS
 from tools.i18n.i18n import I18nAuto
 
 # =======================
-# 配置
+# 模型初始化
 # =======================
 MODEL_DIR = "checkpoints"
 snapshot_download("IndexTeam/IndexTTS-1.5", local_dir=MODEL_DIR)
 
-if not os.path.exists(MODEL_DIR):
-    print(f"Model directory {MODEL_DIR} does not exist. Please download the model first.")
-    sys.exit(1)
-
-for file in [
-    "bigvgan_generator.pth",
-    "bpe.model",
-    "gpt.pth",
-    "config.yaml",
-]:
-    file_path = os.path.join(MODEL_DIR, file)
-    if not os.path.exists(file_path):
-        print(f"Required file {file_path} does not exist. Please download it.")
+required_files = ["bigvgan_generator.pth", "bpe.model", "gpt.pth", "config.yaml"]
+for f in required_files:
+    path = os.path.join(MODEL_DIR, f)
+    if not os.path.exists(path):
+        print(f"Required file {path} not found.")
         sys.exit(1)
 
-# =======================
-# 初始化模型
-# =======================
 i18n = I18nAuto(language="zh_CN")
 tts = IndexTTS(model_dir=MODEL_DIR, cfg_path=os.path.join(MODEL_DIR, "config.yaml"))
 
 os.makedirs("outputs/tasks", exist_ok=True)
 os.makedirs("prompts", exist_ok=True)
 
+# =======================
+# 终端进度回调
+# =======================
+class ProgressPrinter:
+    def __init__(self):
+        self.last = time.time()
+    def __call__(self, progress_value):
+        now = time.time()
+        if now - self.last > 0.5:
+            print(f">> progress: {progress_value*100:.1f}%")
+            self.last = now
+
 # =======================
 # 核心生成函数
 # =======================
@@ -59,17 +58,6 @@ def gen_single(prompt, text, infer_mode,
                max_mel_tokens=600):
     output_path = os.path.join("outputs", f"spk_{int(time.time())}.wav")
     print(">> start inference...")
-
-    # 将进度回调打印到终端
-    class ProgressPrinter:
-        def __init__(self):
-            self.last = time.time()
-        def __call__(self, progress_value):
-            now = time.time()
-            if now - self.last > 0.5:  # 每0.5秒打印一次
-                print(f">> progress: {progress_value*100:.1f}%")
-                self.last = now
-
     tts.gr_progress = ProgressPrinter()
 
     kwargs = {
@@ -108,7 +96,7 @@ def gen_single(prompt, text, infer_mode,
 with gr.Blocks(title="IndexTTS Demo") as demo:
     gr.HTML('''
         <h2><center>IndexTTS: 工业级可控零样本文本转语音系统</h2>
-        <p align="center">(简化版，安全参数，终端打印进度)</p>
+        <p align="center">(简化版，终端打印进度 + 高级参数可调)</p>
     ''')
 
     with gr.Row():
@@ -119,9 +107,22 @@ with gr.Blocks(title="IndexTTS Demo") as demo:
         gen_button = gr.Button("生成语音")
         output_audio = gr.Audio(label="生成结果", visible=True, key="output_audio")
 
+    with gr.Accordion("高级生成参数", open=False):
+        do_sample = gr.Checkbox(label="do_sample", value=True)
+        top_p = gr.Slider(label="top_p", minimum=0.0, maximum=1.0, value=0.9, step=0.01)
+        top_k = gr.Slider(label="top_k", minimum=0, maximum=100, value=50, step=1)
+        temperature = gr.Slider(label="temperature", minimum=0.1, maximum=2.0, value=1.0, step=0.1)
+        length_penalty = gr.Slider(label="length_penalty", minimum=-2.0, maximum=2.0, value=0.0, step=0.1)
+        num_beams = gr.Slider(label="num_beams", minimum=1, maximum=10, value=1, step=1)
+        repetition_penalty = gr.Slider(label="repetition_penalty", minimum=0.1, maximum=20.0, value=1.0, step=0.1)
+        max_mel_tokens = gr.Slider(label="max_mel_tokens", minimum=50, maximum=tts.cfg.gpt.max_mel_tokens,
+                                   value=600, step=10)
+
     gen_button.click(
         gen_single,
-        inputs=[prompt_audio, input_text, infer_mode],
+        inputs=[prompt_audio, input_text, infer_mode,
+                120, 4, do_sample, top_p, top_k, temperature,
+                length_penalty, num_beams, repetition_penalty, max_mel_tokens],
         outputs=[output_audio]
     )