二聚工具箱包装器
pydiarization的Python项目详细描述
pydialization
pydialization是IBDiarization工具包的包装器。
要求
您的路径中必须包含以下二进制文件:
- ffmpeg
- compute mfcc feats和copy feats to htkfrom kaldi
- diarizeme来自ibdialization工具包。
安装
Pydiation可以通过conda安装
conda install -c wdroz pydiarization
或pip
pip install pydiarization
用法
在使用pywrapper之前,必须创建一个包含ibdialization工具包结果的文件夹。
mkdir result.dir
测试安装
若要检查所有二进制文件是否正常工作并被pydialization识别,可以键入:python3 -m pydiarization.test_diarization_wrapper
从代码
将单个视频转换为字符串(.rttm内容)的示例
frompydiarization.diarization_wrapperimportvideo_to_rttm_stringrttm_content=video_to_rttm_string('MY_VIDEO_OR_URL.avi')
下面是所有高级api的列表:
defvideo_to_rttm_string(video_path):""" High-level function that return the rttm as string from a video Arguments: video_path -- where the video is Return: the rttm content as string """defwav_to_rttm_string(wav_path):""" High-level function that return the rttm as string from a wav Arguments: wav_path -- where the wav is Return: the rttm content as string """defrttm_to_string(rttm_path):""" transform rttm file to string Arguments: rttm_path -- where is the rttm file Return: rttm content as string """defrttm_from_video(video_path,rttm_path):""" create a .rttm file from a video Arguments: video_path -- path to the video rttm_path -- path where the .rrtm file will be saved """defrttm_from_wav(wav_path,rttm_path):""" create a .rttm file from a wav Arguments: wav_path -- path to the wav file rttm_path -- path where the .rrtm file will be saved """
来自cli
用法如下:
$ python3 -m pydiarization.run --help usage: run.py [-h] [--video VIDEO] [--wav WAV] [--output OUTPUT] Diarization Toolkit Wrapper CLI optional arguments: -h, --help show this help message and exit --video VIDEO take a video file or URL as input --wav WAV take a wav file as input --output OUTPUT specify to write as .rttm file
一些示例
请求从URL获取.rttm内容:
python3 -m pydiarization.run --video http://data.cstr.inf.ed.ac.uk/summa/data/test.mp4
SPEAKER tmp86hrhwsd 1 0.01 292.17 <NA> <NA> tmp86hrhwsd_spkr_9 <NA> SPEAKER tmp86hrhwsd 1 292.18 2.98 <NA> <NA> tmp86hrhwsd_spkr_2 <NA>
请求改为写入.rttm文件:
python3 -m pydiarization.run --video http://data.cstr.inf.ed.ac.uk/summa/data/test.mp4 --output toto.rttm