Autosub 是用于自动语音识别和字幕生成的实用程序。
但是这个项目已经不再维护了，而且在中国使用有些困难。 于是我克隆了这个项目的代码，在内部把Google Web Speech API替换为了百度短语音识别API 。
项目结构也做出了一些修改，使其即可以在命令行使用，也可以在其他代码中导入使用（主要目的是方便在代码中使用）。语言也改为了仅支持中文。
运行参数有些不一样，要想顺利使用这个新项目你必须提供API_KEY、SECRET_KEY和APP_ID。这里贴出了一些链接，可以跟随这些链接去自行探索：
- 短语音识别标准版产品首页
- 百度AI开放平台语音技术新手入门 （通过阅读可以了解如何获得
API_KEY、SECRET_KEY和APP_ID）
Autosub is a utility for automatic speech recognition and subtitle generation.
But this project is no longer maintained, and it is somewhat difficult to use in China. So I cloned the code of this project and replaced Google Web Speech API with Baidu Short Speech Recognition API internally.
Now Autosub-baidu only supports Chinese.
- 安装 ffmpeg.
- 运行
pip install autosub-baidu.
python -m autosubb -h
usage: __main__.py [-h] [-C CONCURRENCY] [-o OUTPUT] [-F FORMAT] [-L LANG] [-K API_KEY] [-S SECRET_KEY] [-A APP_ID]
[--list-formats] [--list-languages]
[source_path]
positional arguments:
source_path Path to the video or audio file to subtitle
optional arguments:
-h, --help show this help message and exit
-C CONCURRENCY, --concurrency CONCURRENCY
Number of concurrent API requests to make
-o OUTPUT, --output OUTPUT
Output path for subtitles (by default, subtitles are saved in the same directory and name as
the source path)
-F FORMAT, --format FORMAT
Destination subtitle format
-L LANG, --lang LANG Language spoken in source file. default 1537.
-K API_KEY, --api-key API_KEY
The Baidu Cloud API key to be used.
-S SECRET_KEY, --secret-key SECRET_KEY
The Baidu Cloud Secret Key to be used.
-A APP_ID, --app-id APP_ID
The Baidu Cloud AppID to be used.
--list-formats List all available subtitle formats
--list-languages List all available source/destination languages
import sys
from autosubb import generate_subtitles
APP_ID = '2*****0'
API_KEY = 'W********************o'
SECRET_KEY = 'X******************************5'
filename = 'audio.mp3'
def main():
subtitles = generate_subtitles(
source_path=filename,
app_id=APP_ID,
api_key=API_KEY,
secret_key=SECRET_KEY,
concurrency=2,
dev_pid='80001',
)
print(subtitles)
if __name__ == '__main__':
sys.exit(main())[((2.3040000000000003, 4.608000000000002), '从前有一对仙人夫妻。'),
((5.376000000000003, 7.424000000000005), '他们常常到山顶上下棋。'),
...
((56.83200000000004, 59.392000000000046), '猴子的眼睛一直盯着这盘水蜜桃。'),
((60.41600000000005, 63.48800000000005), '战胜对手，其实就是战胜对手的弱点。')]在generate_subtitles函数中指定subtitle_file_format参数：
import sys
from autosubb import generate_subtitles
APP_ID = '2*****0'
API_KEY = 'W********************o'
SECRET_KEY = 'X******************************5'
filename = 'audio.mp3'
def main():
subtitles = generate_subtitles(
source_path=filename,
app_id=APP_ID,
api_key=API_KEY,
secret_key=SECRET_KEY,
concurrency=2,
dev_pid='80001',
subtitle_file_format='srt',
)
with open('audio.srt', 'wb') as out:
out.write(subtitles.encode('utf-8'))
if __name__ == '__main__':
sys.exit(main())如果你想拿到列表形式的字幕（做一些事情），并在稍后保存为SRT文件，
你可以使用srt_formatter函数：
import sys
from autosubb import generate_subtitles
from autosubb import formatters
APP_ID = '2*****0'
API_KEY = 'W********************o'
SECRET_KEY = 'X******************************5'
filename = 'audio.mp3'
def main():
subtitles = generate_subtitles(
source_path=filename,
app_id=APP_ID,
api_key=API_KEY,
secret_key=SECRET_KEY,
concurrency=2,
dev_pid='80001',
)
print(subtitles) # do something
with open('audio.srt', 'wb') as out:
content = formatters.srt_formatter(subtitles)
out.write(content.encode('utf-8'))
if __name__ == '__main__':
sys.exit(main())MIT