Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/vimwei/whispertranscriber
Whisper Transcribe and srt Resegment
https://github.com/vimwei/whispertranscriber
speech-to-text subtitle whisper
Last synced: 3 months ago
JSON representation
Whisper Transcribe and srt Resegment
- Host: GitHub
- URL: https://github.com/vimwei/whispertranscriber
- Owner: VimWei
- License: gpl-3.0
- Created: 2024-10-14T11:40:26.000Z (3 months ago)
- Default Branch: main
- Last Pushed: 2024-10-16T02:50:34.000Z (3 months ago)
- Last Synced: 2024-10-17T02:58:16.851Z (3 months ago)
- Topics: speech-to-text, subtitle, whisper
- Language: Python
- Homepage:
- Size: 40 KB
- Stars: 2
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Whisper Transcribe and SRT-Resegment
## 简介
方便配置 [OpenAI Whisper](https://github.com/openai/whisper) 各种参数,并增强部分功能:
1. 或让Whisper自动断句:
* use_default_line_breaks=True
2. 或自由定义断句的参数:
* use_default_line_breaks=False
* max_line_width=78,
* max_words_per_line=5,
* max_line_count=None,
3. 甚至生成逐字的精准字幕:
* use_default_line_breaks=False
* max_line_width=78,
* max_words_per_line=1,
* max_line_count=None,
4. 甚至实现断句的完全自由调整:
* 首先,(自动)让Whisper自动断句
* 然后,(自由)借助AI批量断句,也可人工任意断句
* 最后,(快速)使用 srt-resegment.py 一键同步精准时间戳
* srt-resegment.py 可独立使用,百搭各种 Whisper 衍生品
* 更有[无缝集成到mpv的lua版本](https://github.com/VimWei/mpv-config)
5. 支持同时处理多个文件
6. 支持同时输出不同格式: SRT/JSON/VTT/TXT## 安装与使用
1. (建议) 新建一个独立的 python 环境,并激活:
* 创建:conda create -n whisper python=3.11
* 激活:conda activate whisper
2. 安装 OpenAI Whisper:
* pip install -U openai-whisper
3. 配置并运行程序:
* python WhisperTranscriber.py
* python srt-resegment.py