通过录音文件识别实现视频自动生成字幕

# 录音文件识别直接支持的格式
SUPPORTED_FORMATS = {'.wav', '.mp3', '.m4a', '.flv', '.mp4', '.wma', '.3gp', '.amr', '.aac', '.ogg', '.flac'}

file_ext = os.path.splitext(input_path)[1].lower()

if file_ext in SUPPORTED_FORMATS:
    # 直接上传，无需转换
    upload_file(input_path, cos_file_path)
else:
    # 不支持的格式，用 ffmpeg 转为 wav
    extract_audio(input_path, audio_path)
    upload_file(audio_path, cos_file_path)

# 使用参数列表 + shell=False，避免路径含空格/特殊字符导致的注入风险
    subprocess.run(
        ["ffmpeg", "-i", video_path, "-vn", "-ar", "16000", audio_path],
        shell=False, check=True,
    )

req = models.CreateRecTaskRequest()
params = {
    "ChannelNum": 1,
    "ResTextFormat": 3,           # 按标点分段，字幕场景专用
    "SentenceMaxLength": 15,      # 单行字幕最大字数
    "SourceType": 0,              # URL 方式
    "ConvertNumMode": 1           # 智能数字转换
}
req._deserialize(params)
req.EngineModelType = "16k_zh"
req.Url = audio_cos_url

resp = client.CreateRecTask(req)
task_id = resp.Data.TaskId

while True:
    resp = client.DescribeTaskStatus(req)
    status = resp.Data.StatusStr
    if status == "success":
        result_detail = resp.Data.ResultDetail
        break
    elif status == "failed":
        print(f"失败: {resp.Data.ErrorMsg}")
        return
    time.sleep(5)

[
    {
        "FinalSentence": "大家好，欢迎来到今天的课程。",
        "StartMs": 500,
        "EndMs": 3200,
        "SpeakerId": 0,
        "SpeechSpeed": 4.2
    },
    {
        "FinalSentence": "今天我们要讲的主题是机器学习。",
        "StartMs": 3500,
        "EndMs": 6800,
        "SpeakerId": 0,
        "SpeechSpeed": 4.5
    }
]

def to_srt(result_detail):
    srt_lines = []
    for i, sentence in enumerate(result_detail, 1):
        start_time = ms_to_srt_time(sentence["StartMs"])
        end_time = ms_to_srt_time(sentence["EndMs"])
        text = sentence["FinalSentence"]
        srt_lines.append(f"{i}\\n{start_time} --> {end_time}\\n{text}\\n")
    return "\\n".join(srt_lines)


def ms_to_srt_time(ms):
    """毫秒转 SRT 时间格式 (HH:MM:SS,mmm)"""
    hours = ms // 3600000
    minutes = (ms % 3600000) // 60000
    seconds = (ms % 60000) // 1000
    millis = ms % 1000
    return f"{hours:02d}:{minutes:02d}:{seconds:02d},{millis:03d}"

1
00:00:00,500 --> 00:00:03,200
大家好，欢迎来到今天的课程。

2
00:00:03,500 --> 00:00:06,800
今天我们要讲的主题是机器学习。

video-srt/
├── main.py              # 主入口：格式判断 + 流程串联
├── common/
│   ├── config.py        # 输出路径配置
│   └── logger.py        # 日志模块
├── cos/
│   ├── config.py        # COS 配置（密钥、桶名、区域）
│   └── cos_cli.py       # COS 客户端（上传、获取URL）
├── ffmpeg/
│   └── ffmpeg.py        # FFmpeg 音频提取（仅不支持格式时使用）
├── tencent/
│   ├── config.py        # ASR 配置（密钥、引擎）
│   └── tencent_cli.py   # ASR 客户端（创建任务、轮询结果）
└── tool/
    ├── time_format.py   # 时间格式化（毫秒→SRT时间）
    └── to_srt.py        # SRT 生成（遍历结果输出字幕）