-
Notifications
You must be signed in to change notification settings - Fork 371
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Intermittent Call Instability After Initial Successful Testing #952
Comments
Which version of agents are you running? |
@BaiMoHan is it still working fine after upgrading your plan? I think it may be related to the issue I brought up in this slack thread. At the bottom of my thread you'll see it seemed to be working fine, but earlier this week it started happening to me again. My experience is the agent leaves the room, and I occasionally see error logs similar to what you see. Here are some that I think are related:
I was on agents version 0.10.1; I'm downgrading to 0.10.0 to see if anything changes |
I have encountered issues with your first and second logs. After upgrading my plan, the logs appear to be normal, but the service is not functioning correctly. I reached out to support, where Mike advised me on adjusting the connection options. However, his recommendations did not prove to be effective. My current issue is that the Speech to Text (STT) function is not consistently working properly. I can hear the user's speech from the Egress, but in the front and backend, the STT does not seem to be running. As of now, I have not received a response from the support team. |
I have developed my agent code based on livekit-examples/voice-pipeline-agent-python and implemented the frontend using livekit/agents-playground. On October 16th, I successfully established a functional call flow. I can confirm that on the morning of October 17th, we conducted multiple successful call tests. However, by the evening of the 17th, I noticed that the call stability had deteriorated.
The issues manifest in the following ways:
Frequent fluctuations where speech is not recognized after a successful connection. The waveforms are visible on the frontend, but there are no changes in the backend logs. Occasionally, after a successful call, the system returns to normal state. I am uncertain if this is related to the livekit.cloud forwarding or any adjustments made.
After a successful room join, approximately 20 seconds later, the logs show "room disconnected."
The rtc_engine emits a warning stating "stream closed," but no other obvious errors are logged.
I have previously raised a few related issues for further reference.
#951
#948
Thank you for your assistance!
The text was updated successfully, but these errors were encountered: