Change Api Model for chatterbox

[IncognitoAnkh]

Hi i was wondering if there is any way to load the multilingual model in the openai tts api extension instead of the default one, i already tryied editing the preset but seems to not work, what i want is not supported? i was trying to do it since the sillytavern implementation dont have a model or language picker so the default model is loaded.

[rsxdalv]

Yes, it's the model_name or similarly named parameter. Are you able to find it?

[IncognitoAnkh]

This is how my preset looks likes.

But maybe is being ignored since my console show "Loading model 'just_a_placeholder'..."
the same model_name as the default preset.

[rsxdalv]

That preset is correct but how do you call the API?

[IncognitoAnkh]

from the tts extension on sillytavern, it has and option for tts webui. maybe is an issue with the extension itself?

[rsxdalv]

There's a language option within the newest version of the extension: https://github.com/rsxdalv/sillytavern-tts-webui

[rsxdalv]

That TTS extension is for direct use of the API, the presets are meant for cases where you can't specify the params, i.e., OpenAI-compatible option

[IncognitoAnkh]

Thank you it worked, but i noticed if you have the option "narrate by paragraphs" enabled, instead of waiting for the next part of the audio all of them overlap each other at the same time. if is disabled it works as intended.

[IncognitoAnkh]

if this helps to found the issue i noticed that if you try to generate an audio while there is already one playing instead of stop the current one and generate the new one (like any other tts source option in sillytavern tts extension), the new audio generate and start to play over the already running making an overlap effect. or directly you get an "network error", if that second thing happen, the only fix is restart tts webui.