BezioBot for Twitch.tv - self hosted TTS solution with self created voice model.

To create voice model I used https://github.com/coqui-ai/TTS using VITS tts model.

• python 3.10

• Requires SoX to exist in ./sox directory or in PATH

• Required curl to exist in root directory or in PATH

• Requires TTS Server 0.13.3 running on http://localhost:5002

• Optionally you can put sounds in .wav format to sounds directory. They will be played using pattern like this [150] sound named 150.wav will be played. Needs to be 22050hz, mono channel.

Eventsub local testing

• Install Twitch CLI
• twitch mock-api generate
  • Grab User ID, Client-ID and Secret and put it in config
• twitch mock-api start
• twitch event websocket start-server -p 4000
• Install all requirements of twitch_tts_bot
• Start TTS server tts-server --model_path best_model.pth --config_path config.json --use_cuda true
• Trigger custom reward event twitch event trigger channel.channel_points_custom_reward_redemption.add -T websocket -t <USER_ID> -u <SUBSCRIPTION_ID>