Repository navigation
Configuration
Most if not all settings for the Twitch bot will update immediately when you click save, including any active commands. If they do not, a restart of the program will resolve any issues.
Your channel name that will run the chat bot and request authorization from. This is autopupulated once you have authorized the app to use your login.
Choose between !, ~, +, & as command prefixes.
See Commands for more information
- Join
- Speak
- Set Voice
- List Voices
- Help
The help command and all cooldowns are currently only configurable in the config.ini generated after first launch. To configure these settings not available in the UI, please do so when the application is not running.
Under browser source, you can copy the current URL for the browser source to use in OBS. [default http://localhost:5000/overlay]
Additionally, you can set the current port of the local webserver [default 5000]. If you change this port, please restart the application to ensure everything works as expected.
Your Elevenlabs API key from your account. See Elevenlabs Integration for more details.
The number of credits remaining which if you get below, the program will warn you about.
The list of all currently added 11L voices. You can click on one to select it to either remove or preview.
Import all of your 11L voices from your account at once.
Add a specific 11L voice (from your account) to the list of added voices. Requires its voice-id.
Remove the currently selected voice. All members with that voice set as their TTS voice will be reset to a Local TTS voice.
When a voice is selected, clicking this will play a preview voice clip of the voice.
Previewing a voice will do a one-time fetch using credits, but will cache voices for a few months if you wish to preview a specific voice again.
Pocket TTS is a locally running small TTS model. It's under 100M params which lets you run it very fast on relatively accessible hardware like most modern desktop CPUs. Even at this size, it supports voice cloning from a WAV audio sample.
- First, download the model from the huggingface repository. The model is gated, so you need to log in or create an account with huggingface and then get authorized to actually download it. The model should be named
tts_b6369a24.safetensorsin the files tab.
To get authorized, simply put in your usage company/usage as for fun/hobby on the pocket-tts page.
Direct link to model (make sure you do NOT download the one without voice cloning)
- In the Settings tab under Pocket TTS section, click
Select Modelbutton. This opens up a file selection dialog. Pick the previously downloaded model file. Pocket TTS will be configured now. You can proceed to add voices to use the model.
-
First, make an audio sample of the voice you want to clone. This sample, preferably should be under 10 seconds in length, and it's file size should preferably be less than 5MB. It MUST be in WAV file format.
-
Click on
Add Voicebutton on the Pocket TTS section. This should open up an Add new Pocket Voice dialog. -
In the dialog, enter an appropriate name for the voice, and then click on
Pick WAV Voice File to Clone. This should open up a file picker dialog. -
Pick the previously sourced sample audio, and click
Add. This should create a new entry inside the voices list with your newly created audio. You can now preview or use the newly configured voice.
Add a voice to the list of added voices by choosing a WAV audio sample to clone.
Remove the currently selected voice. All members with that voice set as their TTS voice will be reset to a Local TTS voice.
When a voice is selected, clicking this will play a preview voice clip of the voice.