Upload app.py with huggingface_hub

app.py

@@ -1,15 +1,19 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
 import os
 import sys
 import time
-import threading
+import json
+import argparse
 import warnings
 
 from huggingface_hub import snapshot_download
-import gradio as gr
 
 warnings.filterwarnings("ignore", category=FutureWarning)
 warnings.filterwarnings("ignore", category=UserWarning)
 
+# 添加本地模块路径
 current_dir = os.path.dirname(os.path.abspath(__file__))
 sys.path.append(current_dir)
 sys.path.append(os.path.join(current_dir, "indextts"))
@@ -17,48 +21,78 @@ sys.path.append(os.path.join(current_dir, "indextts"))
 from indextts.infer import IndexTTS
 from tools.i18n.i18n import I18nAuto
 
-# =======================
-# 模型初始化
-# =======================
-MODEL_DIR = "checkpoints"
-snapshot_download("IndexTeam/IndexTTS-1.5", local_dir=MODEL_DIR)
-
+# ---------------------------
+# 参数解析
+# ---------------------------
+parser = argparse.ArgumentParser(description="IndexTTS CLI")
+parser.add_argument("--model_dir", type=str, default="checkpoints", help="模型目录")
+parser.add_argument("--output_dir", type=str, default="outputs", help="音频输出目录")
+parser.add_argument("--verbose", action="store_true", help="打印详细信息")
+cmd_args = parser.parse_args()
+
+os.makedirs(cmd_args.model_dir, exist_ok=True)
+os.makedirs(cmd_args.output_dir, exist_ok=True)
+
+# ---------------------------
+# 下载模型（如果不存在）
+# ---------------------------
+print(">> 检查模型...")
+snapshot_download(
+    repo_id="IndexTeam/IndexTTS-1.5",
+    local_dir=cmd_args.model_dir,
+    local_dir_use_symlinks=False,
+)
 required_files = ["bigvgan_generator.pth", "bpe.model", "gpt.pth", "config.yaml"]
 for f in required_files:
-    path = os.path.join(MODEL_DIR, f)
+    path = os.path.join(cmd_args.model_dir, f)
     if not os.path.exists(path):
-        print(f"Required file {path} not found.")
-        sys.exit(1)
-
-i18n = I18nAuto(language="zh_CN")
-tts = IndexTTS(model_dir=MODEL_DIR, cfg_path=os.path.join(MODEL_DIR, "config.yaml"))
+        raise FileNotFoundError(f"缺少模型文件: {path}")
 
-os.makedirs("outputs/tasks", exist_ok=True)
-os.makedirs("prompts", exist_ok=True)
+print(">> 模型文件检查完成")
 
-# =======================
-# 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
-# =======================
-def gen_single(prompt, text, infer_mode,
-               max_text_tokens_per_sentence=120, sentences_bucket_max_size=4,
-               do_sample=True, top_p=0.9, top_k=50, temperature=1.0,
-               length_penalty=0.0, num_beams=1, repetition_penalty=1.0,
-               max_mel_tokens=600):
-    output_path = os.path.join("outputs", f"spk_{int(time.time())}.wav")
-    print(">> start inference...")
-    tts.gr_progress = ProgressPrinter()
+# ---------------------------
+# 初始化 TTS
+# ---------------------------
+i18n = I18nAuto(language="zh_CN")
+tts = IndexTTS(model_dir=cmd_args.model_dir, cfg_path=os.path.join(cmd_args.model_dir, "config.yaml"))
+
+# ---------------------------
+# 语音生成函数
+# ---------------------------
+def generate_audio(
+    text,
+    output_path=None,
+    prompt_audio=None,
+    infer_mode="普通推理",
+    max_text_tokens_per_sentence=120,
+    sentences_bucket_max_size=4,
+    do_sample=True,
+    top_p=0.8,
+    top_k=30,
+    temperature=1.0,
+    length_penalty=0.0,
+    num_beams=3,
+    repetition_penalty=10.0,
+    max_mel_tokens=None,
+    verbose=True
+):
+    """
+    Args:
+        text: 要生成的文本
+        output_path: 输出 wav 文件路径
+        prompt_audio: 参考音频路径
+        infer_mode: "普通推理" 或 "批次推理"
+        max_text_tokens_per_sentence: 分句最大 token 数
+        sentences_bucket_max_size: 分句分桶最大容量（批次模式有效）
+        do_sample, top_p, top_k, temperature, length_penalty, num_beams, repetition_penalty, max_mel_tokens:
+            TTS 模型生成参数
+        verbose: 是否打印进度
+    """
+    if output_path is None:
+        output_path = os.path.join(cmd_args.output_dir, f"spk_{int(time.time())}.wav")
+
+    if max_mel_tokens is None:
+        max_mel_tokens = tts.cfg.gpt.max_mel_tokens
 
     kwargs = {
         "do_sample": bool(do_sample),
@@ -71,66 +105,37 @@ def gen_single(prompt, text, infer_mode,
         "max_mel_tokens": int(max_mel_tokens),
     }
 
+    print(">> start inference...")
+    start_time = time.time()
     try:
         if infer_mode == "普通推理":
-            tts.infer(prompt, text, output_path,
-                      verbose=True,
-                      max_text_tokens_per_sentence=int(max_text_tokens_per_sentence),
-                      **kwargs)
+            tts.gr_progress = None
+            tts.infer(
+                prompt_audio, text, output_path,
+                verbose=verbose,
+                max_text_tokens_per_sentence=int(max_text_tokens_per_sentence),
+                **kwargs
+            )
         else:
-            tts.infer_fast(prompt, text, output_path,
-                           verbose=True,
-                           max_text_tokens_per_sentence=int(max_text_tokens_per_sentence),
-                           sentences_bucket_max_size=int(sentences_bucket_max_size),
-                           **kwargs)
+            tts.gr_progress = None
+            tts.infer_fast(
+                prompt_audio, text, output_path,
+                verbose=verbose,
+                max_text_tokens_per_sentence=int(max_text_tokens_per_sentence),
+                sentences_bucket_max_size=int(sentences_bucket_max_size),
+                **kwargs
+            )
+        print(f">> wav file saved to: {output_path}")
+        print(f">> total time: {time.time() - start_time:.2f}s")
     except Exception as e:
-        print(f">> Error during inference: {e}")
-        return None
-
-    print(f">> wav file saved to: {output_path}")
+        print(">> 生成失败:", e)
+        raise e
     return output_path
 
-# =======================
-# Gradio界面
-# =======================
-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，终端打印进度 + 高级参数可调)</p>
-    ''')
-
-    with gr.Row():
-        prompt_audio = gr.Audio(label="参考音频", key="prompt_audio",
-                                sources=["upload","microphone"], type="filepath")
-        input_text = gr.TextArea(label="文本", key="input_text", placeholder="请输入目标文本")
-        infer_mode = gr.Radio(choices=["普通推理", "批次推理"], label="推理模式", value="普通推理")
-        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,
-                120, 4, do_sample, top_p, top_k, temperature,
-                length_penalty, num_beams, repetition_penalty, max_mel_tokens],
-        outputs=[output_audio]
-    )
-
-# =======================
-# 启动
-# =======================
-def main():
-    demo.launch(server_name="0.0.0.0", server_port=7860)
-
+# ---------------------------
+# CLI 测试
+# ---------------------------
 if __name__ == "__main__":
-    main()
+    sample_text = "你好，欢迎使用 IndexTTS。"
+    output_file = generate_audio(sample_text)
+    print(f">> 生成完成: {output_file}")