Very helpful tutorial, i already made my own assistant for uni assignment but was struggling to connect it to vtube studio. That said this only works windows for mac its alot more painfull as you can't assign specific devices to specific apps
this uses silero to generate a voice, for it to be changed going into documents of silero tts can help change the voice u want, if you want to get a more accurate change of voice you want i'd look into rvc pipeline since with rvc u can train voice models into what u want the voice to be and there are ton of voice models out there with rvc.
what are you using locally? Oobabooga Webui ? Ollama? LM Studio? , Usually every Local Api address changes per “back end” program you’re using , for example LM Studio is defualport is “localhost:1234/v1” or “0.0.0.0:1234/v1” and for ooba webui defualt address is “0.0.0.0:5000/v1”
This program uses Silero TTS, if silero has korean voice models more than likely, and also to transcribe Voice to text it uses whisper openai, You may need a fine tuned Korean model,
@@i6odoh thx I've decided on the Korean model. I'm planning to use the saltlux/Ko-Llama3-Luxia-8B model. So, are there any other processes or additional processes that need to be added??
![How to make an AI Vtuber [instillation]](http://i.ytimg.com/vi/iiDnzmQ5008/mqdefault.jpg)
![I.N "HALLUCINATION" | [Stray Kids : SKZ-PLAYER]](http://i.ytimg.com/vi/n5B5q1Hwt_U/mqdefault.jpg)
Very helpful tutorial, i already made my own assistant for uni assignment but was struggling to connect it to vtube studio. That said this only works windows for mac its alot more painfull as you can't assign specific devices to specific apps
Glad to know , I’ve never used Mac OS, just Linux and Windows so far,
This is really cool, is there a way to change the voice of the avatar? If so, can you somehow add custom voicemodels into it?
this uses silero to generate a voice, for it to be changed going into documents of silero tts can help change the voice u want, if you want to get a more accurate change of voice you want i'd look into rvc pipeline since with rvc u can train voice models into what u want the voice to be and there are ton of voice models out there with rvc.
Beautiful!
I'm getting some "max retries exceeded with url..." error... do you know any fix for that?
That seems to be that your ip address is wrong in the “Chatbot.py” file
what are you using locally? Oobabooga Webui ? Ollama? LM Studio? , Usually every Local Api address changes per “back end” program you’re using , for example LM Studio is defualport is “localhost:1234/v1” or “0.0.0.0:1234/v1” and for ooba webui defualt address is “0.0.0.0:5000/v1”
I want to make it work in Korean, but is there any way?
This program uses Silero TTS, if silero has korean voice models more than likely, and also to transcribe Voice to text it uses whisper openai, You may need a fine tuned Korean model,
@@i6odoh thx I've decided on the Korean model. I'm planning to use the saltlux/Ko-Llama3-Luxia-8B model. So, are there any other processes or additional processes that need to be added??
Not sure yet i’m sure some googling around github repo’s or/and hugging face there should be fine tuned models,
@@i6od ah, so you think it would be difficult to modify this model??
@@i6od sir, so much thank you I'll be back.