Skip to content

Whisper based Japanese subtitle generator

License

Notifications You must be signed in to change notification settings

InsWan/N46Whisper

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

N46Whisper

Language : English | 简体中文

N46Whisper is a Google Colab notebook application that developed for streamlined video subtitle file generation to improve productivity of Nogizaka46 (and Sakamichi groups) subbers.

The notebook is based on Whisper, a general-prupose speech recognition model.

The output file will be in Advanced SubStation Alpha(ass) format with built-in style of selected sub group so it can be directly imported into Aegisub for subsequent editing.

What's New:

2023.3.15:

  • Add functions to split multiple words/sententces in one line.
  • Update documents and other minor fixes.

2023.3.12:

  • Add chatGPT translation and bilingual subtitle file generation features.
  • Update documents and other minor fixes.

2023.01.26:

  • Update scripts to reflect recent changes in Whisper.

2022.12.31:

  • Allow user to select files directly from mounted google drive.
  • Other minor fixes.

How to use

  • Click here to open the notebook in Google Colab.
  • Upload file and follow the instruction to run the notebook.
  • The ass file will be automatically downloaded once done.

AI translation

The notebook now allow users to translate transcribed subtitle text line by line using AT translation tools.

Currently, it supports chatGPT and the default target language is zh-CN.

The translated text will be append in the same line after the original text and sepearted by /N, such that a new bilingual subtitle file is generated.

For instance:

QQ截图20230312155700

An example of bilingual subtitle:

QQ截图20230312160015

To use the AI translation, users must use their own OpenAI API Key. To obtain a free Key, go to https://platform.openai.com/account/api-keys

Please note there will be limitaions on usage for free keys, choose a paid plan to speed up at your own cost.

Split lines

Users can choose to split text in a single line by space.The child lines will have same time stamp with the parent line, respectively.

For instance, for a line contains multiple long sentences:

Dialogue: 0,0:01:00.52,0:01:17.52,default,,0,0,0,,Birthday Liveについて話そうかなと思います よろしくお願いします

After split:

Dialogue: 0,0:01:00.52,0:01:17.52,default,,0,0,0,,Birthday Liveについて話そうかなと思います(adjust_required)

Dialogue: 0,0:01:00.52,0:01:17.52,default,,0,0,0,,ろしくお願いします(adjust_required)

Support

The application could significantly reduce the labour and time costs of sub-groups or individual subbers. However, despite its impressive performance, the Whisper model and the application itself are not without limitations.Please read the orgininal documents and Discussions to learn more about the usage of Whisper and the common issues.

However, if you have any throughts, requests or questions that directly related to making subtitiles for Sakamichi group girls, please feel free to post here or contact me

License

The code is released under the MIT license. See License for details.

About

Whisper based Japanese subtitle generator

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Jupyter Notebook 67.8%
  • Python 32.2%