main.py

#!/usr/bin/env python3
"""
ezA2T - 实时语音转字幕软件主程序
Easy Audio to Text - Real-time Speech to Subtitle Application
"""

import sys
import os
import threading
import time
import argparse
from pathlib import Path

# 添加src目录到路径
sys.path.insert(0, os.path.join(os.path.dirname(__file__), 'src'))

class RealTimeSubtitleApp:
    """实时字幕应用主类"""
    
    def __init__(self, config_file: str = "config.json"):
        # 延迟导入模块
        from src.config import Config
        
        self.config = Config(config_file)
        self.audio_capture = None
        self.speech_recognizer = None
        self.subtitle_window = None
        self.is_running = False
        
        print("ezA2T - 实时语音转字幕软件")
        print("=" * 40)
        
    def initialize_components(self):
        """初始化各个组件"""
        # 延迟导入模块
        from src.wasapi_loopback_capture import WASAPILoopbackCapture
        from src.speech_recognizer import SpeechRecognizer, SimpleSpeechRecognizer, VOSK_AVAILABLE
        from src.subtitle_ui import SubtitleWindow
        
        print("正在初始化组件...")
        
        # 初始化WASAPI loopback音频捕获（专门捕获电脑播放的音频）
        try:
            sample_rate = self.config.get('audio.sample_rate', 16000)
            chunk_size = self.config.get('audio.chunk_size', 1024)
            
            print("初始化WASAPI loopback音频捕获...")
            self.audio_capture = WASAPILoopbackCapture(sample_rate, chunk_size)
            
            # 列出可用设备
            print("\\n可用音频设备:")
            self.audio_capture.list_all_devices()
            
        except Exception as e:
            print(f"音频捕获初始化失败: {e}")
            return False
        
        # 初始化语音识别
        print("\\n初始化语音识别...")
        model_path = self.config.get('recognition.model_path', './models/vosk-model-cn-0.22')
        
        if VOSK_AVAILABLE:
            self.speech_recognizer = SpeechRecognizer(model_path, sample_rate)
            if not self.speech_recognizer.load_model():
                print("VOSK模型加载失败，请检查模型路径或下载模型")
                print("\\n模型下载说明:")
                print("1. 访问 https://alphacephei.com/vosk/models")
                print("2. 下载中文模型（推荐 vosk-model-small-cn-0.22）")
                print("3. 解压到 models/ 目录下")
                
                # 使用简单识别器作为备用
                print("\\n使用模拟识别器（仅用于测试界面）")
                self.speech_recognizer = SimpleSpeechRecognizer()
        else:
            print("VOSK未安装，使用模拟识别器")
            self.speech_recognizer = SimpleSpeechRecognizer()
        
        # 初始化字幕界面
        print("\\n初始化字幕界面...")
        self.subtitle_window = SubtitleWindow(self.config)
        
        print("组件初始化完成！")
        return True
    
    def start_recognition(self):
        """开始语音识别"""
        if not self.audio_capture or not self.speech_recognizer:
            print("组件未正确初始化")
            return False
        
        print("\\n开始实时语音识别...")
        self.is_running = True
        
        # 开始语音识别
        self.speech_recognizer.start_recognition(self.on_recognition_result)
        
        # 开始音频捕获
        if not self.audio_capture.start_recording(self.on_audio_data):
            print("无法开始捕获音频")
            return False
        
        print("语音识别已启动")
        print("提示:")
        print("- 右键点击字幕窗口可以打开设置菜单")
        print("- 双击字幕窗口可以切换置顶状态")
        print("- 拖拽字幕窗口可以移动位置")
        print("- 按 Ctrl+S 保存字幕，Ctrl+Q 或 ESC 退出")
        
        return True
    
    def stop_recognition(self):
        """停止语音识别"""
        if not self.is_running:
            return
        
        print("\\n正在停止语音识别...")
        self.is_running = False
        
        if self.audio_capture:
            self.audio_capture.stop_recording()
        
        if self.speech_recognizer:
            self.speech_recognizer.stop_recognition()
        
        print("语音识别已停止")
    
    def on_audio_data(self, audio_data):
        """音频数据回调"""
        if self.is_running and self.speech_recognizer:
            self.speech_recognizer.process_audio(audio_data)
    
    def on_recognition_result(self, text: str, is_final: bool = True):
        """语音识别结果回调"""
        if self.subtitle_window:
            self.subtitle_window.update_subtitle(text, is_final)
        
        # 打印识别结果到控制台
        if is_final and text.strip():
            print(f"识别结果: {text}")
    
    def run(self):
        """运行应用"""
        try:
            # 初始化组件
            if not self.initialize_components():
                print("初始化失败")
                return
            
            # 开始识别
            if not self.start_recognition():
                print("启动识别失败")
                return
            
            # 运行界面（阻塞）
            self.subtitle_window.run()
            
        except KeyboardInterrupt:
            print("\\n用户中断")
        except Exception as e:
            print(f"运行出错: {e}")
        finally:
            self.stop_recognition()
            print("程序已退出")

def download_model_guide():
    """显示模型下载指南"""
    print("\\nezA2T 模型下载指南")
    print("=" * 30)
    print("\\n为了使用离线语音识别，您需要下载VOSK模型:")
    print("\\n1. 访问官方模型库:")
    print("   https://alphacephei.com/vosk/models")
    print("\\n2. 推荐的中文模型:")
    print("   - vosk-model-small-cn-0.22 (约40MB，适合一般使用)")
    print("   - vosk-model-cn-0.22 (约1.2GB，更高精度)")
    print("\\n3. 英文模型:")
    print("   - vosk-model-small-en-us-0.15 (约40MB)")
    print("   - vosk-model-en-us-0.22 (约1.8GB)")
    print("\\n4. 下载后解压到项目的 models/ 目录下")
    print("\\n5. 修改 config.json 中的 model_path 指向对应模型")
    print("\\n示例目录结构:")
    print("ezA2T/")
    print("├── models/")
    print("│   ├── vosk-model-small-cn-0.22/")
    print("│   │   ├── am/")
    print("│   │   ├── graph/")
    print("│   │   └── ...")
    print("│   └── vosk-model-small-en-us-0.15/")
    print("└── ...")

def check_dependencies():
    """检查依赖项"""
    missing_deps = []
    
    try:
        import soundcard
    except ImportError:
        missing_deps.append('soundcard')
    
    try:
        import vosk
    except ImportError:
        missing_deps.append('vosk')
    
    try:
        import numpy
    except ImportError:
        missing_deps.append('numpy')
    
    if missing_deps:
        print("缺少以下依赖项:")
        for dep in missing_deps:
            print(f"  - {dep}")
        print("\\n请运行以下命令安装:")
        print(f"pip install {' '.join(missing_deps)}")
        return False
    
    return True

def main():
    """主函数"""
    parser = argparse.ArgumentParser(description='ezA2T - 实时语音转字幕软件')
    parser.add_argument('--config', '-c', default='config.json', help='配置文件路径')
    parser.add_argument('--check-deps', action='store_true', help='检查依赖项')
    parser.add_argument('--download-guide', action='store_true', help='显示模型下载指南')
    parser.add_argument('--list-devices', action='store_true', help='列出音频设备')
    
    args = parser.parse_args()
    
    if args.check_deps:
        if check_dependencies():
            print("所有依赖项已安装")
        return
    
    if args.download_guide:
        download_model_guide()
        return
    
    if args.list_devices:
        try:
            from src.wasapi_loopback_capture import WASAPILoopbackCapture
            audio = WASAPILoopbackCapture()
            audio.list_all_devices()
        except Exception as e:
            print(f"列出设备失败: {e}")
        return
    
    # 检查依赖项
    if not check_dependencies():
        return
    
    # 创建并运行应用
    app = RealTimeSubtitleApp(args.config)
    app.run()

if __name__ == "__main__":
    main()