HI, I created a GPT3 powered AI language partner which allows you to practice your oral language. The app allows you to set the specific context of the conversation and listen to the respond. You can try it out here: APP
Now through testing, I have found some problems.
- the google stun server cannot be used in China region, where they complain that the webrtc component always says “No frame arrives”
- Now I use azure service for tts and stt. I find that I have to save the audio file after the speaker is finished, then resend the audio file to azure service which is quite slow.
Do you guys know how to fix these issues. here is the source code: source code