pip install openai-whisper

import whisper
import os
from moviepy.editor import VideoFileClip
from datetime import timedelta
import torch

def extract_audio(video_path, audio_path):
    """從視頻中提取音頻"""
    try:
        video = VideoFileClip(video_path)
        video.audio.write_audiofile(audio_path)
        video.close()
    except Exception as e:
        print(f"音頻提取錯(cuò)誤: {str(e)}")
        raise

def generate_srt(segments, output_srt):
    """生成SRT格式字幕文件"""
    with open(output_srt, 'w', encoding='utf-8') as f:
        for i, segment in enumerate(segments, start=1):
            # 轉(zhuǎn)換時(shí)間格式
            start = str(timedelta(seconds=int(segment['start']))) + ',000'
            end = str(timedelta(seconds=int(segment['end']))) + ',000'
            
            # 寫入SRT格式
            f.write(f"{i}\n")
            f.write(f"{start} --> {end}\n")
            f.write(f"{segment['text'].strip()}\n\n")

def main(video_path, output_srt):
    """主函數(shù)"""
    try:
        # 檢查CUDA是否可用
        device = "cuda" if torch.cuda.is_available() else "cpu"
        print(f"使用設(shè)備: {device}")

        # 加載模型
        print("加載Whisper模型...")
        model = whisper.load_model("base", device=device)

        # 直接使用視頻文件進(jìn)行轉(zhuǎn)錄
        print("開始轉(zhuǎn)錄...")
        result = model.transcribe(video_path, language="zh")

        # 生成字幕文件
        print("生成字幕文件...")
        generate_srt(result["segments"], output_srt)

        print(f"字幕已生成: {output_srt}")

    except Exception as e:
        print(f"處理過程中出錯(cuò): {str(e)}")
        raise

if __name__ == "__main__":
    # 設(shè)置輸入輸出路徑
    video_path = "your_video.mp4"  # 替換為你的視頻文件路徑
    output_srt = "output.srt"      # 輸出的字幕文件路徑
    
    main(video_path, output_srt)

print("加載Whisper模型...")
model = whisper.load_model("base", device=device)

pip install torch torchvision torchaudio
pip install moviepy

from tqdm import tqdm

def generate_srt_with_progress(segments, output_srt):
    """帶進(jìn)度顯示的字幕生成"""
    with open(output_srt, 'w', encoding='utf-8') as f:
        for i, segment in tqdm(enumerate(segments, start=1), 
                             desc="生成字幕", 
                             total=len(segments)):
            start = str(timedelta(seconds=int(segment['start']))) + ',000'
            end = str(timedelta(seconds=int(segment['end']))) + ',000'
            
            f.write(f"{i}\n")
            f.write(f"{start} --> {end}\n")
            f.write(f"{segment['text'].strip()}\n\n")

import logging
import sys

# 配置日志
logging.basicConfig(
    level=logging.INFO,
    format='%(asctime)s - %(levelname)s - %(message)s',
    handlers=[
        logging.FileHandler('subtitle_generation.log'),
        logging.StreamHandler(sys.stdout)
    ]
)

def main(video_path, output_srt):
    """帶有完整錯(cuò)誤處理和日志的主函數(shù)"""
    try:
        if not os.path.exists(video_path):
            raise FileNotFoundError(f"視頻文件不存在: {video_path}")

        logging.info(f"開始處理視頻: {video_path}")
        
        # 檢查CUDA
        device = "cuda" if torch.cuda.is_available() else "cpu"
        logging.info(f"使用設(shè)備: {device}")

        # 加載模型
        logging.info("加載Whisper模型...")
        model = whisper.load_model("base", device=device)

        # 轉(zhuǎn)錄
        logging.info("開始轉(zhuǎn)錄...")
        result = model.transcribe(video_path, language="zh")

        # 生成字幕
        logging.info("生成字幕文件...")
        generate_srt_with_progress(result["segments"], output_srt)

        logging.info(f"字幕生成完成: {output_srt}")

    except FileNotFoundError as e:
        logging.error(f"文件錯(cuò)誤: {str(e)}")
        raise
    except Exception as e:
        logging.error(f"處理錯(cuò)誤: {str(e)}")
        raise

def process_long_video(video_path, output_srt, segment_duration=300):
    """分段處理長(zhǎng)視頻"""
    from moviepy.editor import VideoFileClip
    
    video = VideoFileClip(video_path)
    duration = video.duration
    segments = []
    
    for start in range(0, int(duration), segment_duration):
        end = min(start + segment_duration, duration)
        
        # 提取片段
        segment = video.subclip(start, end)
        temp_audio = f"temp_{start}_{end}.wav"
        segment.audio.write_audiofile(temp_audio)
        
        # 處理片段
        result = model.transcribe(temp_audio, language="zh")
        segments.extend(result["segments"])
        
        # 清理臨時(shí)文件
        os.remove(temp_audio)
    
    # 生成完整字幕
    generate_srt(segments, output_srt)
    video.close()

import whisper

# 加載模型
model = whisper.load_model("base")

# 設(shè)置視頻或音頻文件路徑
audio_path = "path_to_audio_or_video"

# 進(jìn)行語音翻譯，英文語音翻譯成中文
result = model.transcribe(audio_path, language="en", task="translate")

# 打印翻譯結(jié)果
print(result["text"])