Real-time Dialogue Intermittent Issues During Conversation

gpt-omni / mini-omni2

Towards Open-source GPT-4o with Vision, Speech and Duplex Capabilities。

https://arxiv.org/abs/2410.11190

MIT License

1.55k stars 183 forks source link

Real-time Dialogue Intermittent Issues During Conversation #43

Open crzwork2024 opened 2 days ago

crzwork2024 commented 2 days ago

Great project.

The installation process in Windows went smoothly for me, but during conversations, the real-time dialogue is intermittent, while the manually generated dialogue is very smooth. Does anyone know the reason?

Thanks!

mini-omni commented 1 day ago

did you use GPU or CPU? how many tokens per sec in server log?

crzwork2024 commented 23 hours ago

Hi, I was using GPU, and where to check the number of tokens per sec? I basically did nothing and just was following the github instruction, nothing changed in your original code. Are there any parameters I should play around in order to work smoothly? thanks!

crzwork2024 commented 23 hours ago

I can add that the input was always interuppted around 2 seconds, not sure why