Skip to content

mlof/Whippersnapper

Repository files navigation

Whippersnapper

A small transcription bot for Discord.

Usage

  1. Create a Discord bot in the developer portal.
  2. Add the bot's token to appsettings.json.
  3. Go to the OAuth2 -> Url Generator tab in the Discord Developer Portal, select the bot scope, ensure the "Send Messages" and "Embed Links" permissions are selected and copy the generated URL.
  4. Paste the URL into your browser and select the server you want to add the bot to.
  5. Run the bot.

It will then automatically transcribe any voice messages sent in the server.

Configuration

The following configuration options are available:

Option Description Default
BotToken The Discord bot token. ""
ModelFile The path to the model file. "ggml-base.bin"
KeepAttachments Whether to keep the audio files after transcription. false
StatusMessage The message to display as the bot's status. "Always listening, never sleeping."
BadWords A list of words to censor. ["PHP"]
Translate Whether to translate the transcription to English. false
Strategy The strategy to use for translation. "WHISPER_SAMPLING_BEAM_SEARCH"
Threads The number of threads to use for transcription. 4

Strategy

The strategy option determines how the translation is performed. The following strategies are available:

Strategy Description
WHISPER_SAMPLING_GREEDY Greedy sampling.
WHISPER_SAMPLING_BEAM_SEARCH Beam search sampling.

This can be configured in appsettings.json. If you want to use a different model, you can download one from here but i've found that the default one works well enough.

About

A small transcription bot for Discord.

Resources

Stars

Watchers

Forks

Packages

 
 
 

Languages