You can think of it as voice controlled soundboard
Bot listens -> Vosk speech to text model transcribes -> Bot plays a particular song if certain words are heard.
Invite the bot using the link
Use /help command to see what you can do.
Only english, turkish and dutch is supported. Contact me for further language support.
!! A bit outdated. I will update with more details later.
Install docker composer and use docker compose up! Don't forget to set DISCORD_TOKEN, DATABASE_URL and OWNER_ID env variables.
Only external dependency you need is Opus codec that discord uses. If you are on linux/Mac, You can get it from your package manager. You need to manually build it on windows. Read the original songbird repo for more info.
After ensuring that can run it, to actually run it, change .env.example into .env, add your discord bot token, then just cargo install just and then type just build and just run. Otherwise, follow this repo
Manually copying lib files to target dir makes cargo recompile it all. I might fix that in future.
TODO:
- Make dev experience better (build script and copy without recompile)
- Checking the integrity of the audio files
- Compress files before storing
- Better audio list with buttons etc
- Preview
Future Plans:
- Website for uploading and managing sounds.
I made this to get a laugh from my friends while watching f1 together. It used to support only one prompt. I made it so that the bot played "dududu max verstappen" when it head "verstappen". Everyone liked the idea so much, so I made it into a bot that is usable by everyone.