Akash98Sky / gemi-bot

Gemi is a telegram chat bot assistant with text, image and file handling capabilities.
https://t.me/MyGemibot
MIT License
1 stars 4 forks source link

Voice not working properly #9

Closed Deplyapp closed 6 months ago

Deplyapp commented 6 months ago

I have tried using the voice feature, but it doesn't work properly. Sometimes, when I say "What is 2+2" in the voice message, the bot doesn't understand it correctly. Additionally, it doesn't send voice messages when I try to send them.

Akash98Sky commented 6 months ago

I am using a free service to run this bot, which has a very minimal system specification. As a result, I can only run some light voice recognition models. So this is expected.

If you want to run it yourself feel free to do that. Here's the code for how a vosk model is loaded: https://github.com/Akash98Sky/gemi-voice/blob/main/stt/vosk.py#L14 You can download vosk models from here: https://alphacephei.com/vosk/models

But I will keep your feedback in mind for future improvements.

Additionally, it doesn't send voice messages when I try to send them.

This depends on the question you ask. Or, if you ask it to reply to it over voice or audio it'll do the same.

Deplyapp commented 6 months ago

Thanks for the guidance