Skip to content

An application for editing audio-text datasets with ease

Notifications You must be signed in to change notification settings

hopto-dot/VoiceDatasetEditor

Repository files navigation

VoiceDatasetEditor

日本語はこちら

Voice Dataset Editor is an application for editing audio-text datasets with ease and has been localised into Japanese.

Latest release here

User interface

Voice Dataset Editor supports .list files in the format:

audio_file_name.wav|speaker_name|language|transcription

For example, the following file is valid:

voice_0-0.wav|Hestia|JP|へへ、心配してくれるのかい?
voice_0-1.wav|Hestia|JP|僕は今、べらぼうにはしゃぎ回っている!
voice_1.wav|Hestia|JP|炎は燃え続け、照らし続ける!

Features

  • Drag and drop .list files into the application to load, OR, go to [File > Load dataset] then find and select the .list file
  • Edit lines and click save next to the transcription to save individual transcriptions
  • Save all transcriptions on the current page with the Save button at the top
  • Change how many items are shown per page in the settings
  • Application can be displayed in English or Japanese
  • The program is automatically displayed in Japanese if it detects your system language is Japanese
  • Search and replace
  • Sort dataset entries by file name, transcription or audio length
  • Move audio files that have no corresponding transcription out of the audio folder
  • Remove transcriptions based on conditions (audio length, contains text etc.)
  • Trim leading and trailing characters
  • Add display settings such as font size, item width and more
  • Possible support for other display languages (if in demand)

About

An application for editing audio-text datasets with ease

Resources

Stars

Watchers

Forks

Languages