We have been using the scale subscription that allows 15 concurrent devices to run at the same time. And this issue happen often during our live training sessions. The avatar auto starts the conversation and we have have used narrative design Trigger to start the conversation by the avatar.
And sometimes it throws this error and the avatar does not say anything.
[Error][Character]: Grpc.Core.RpcException: Status(StatusCode=Unknown, Detail=“Unable to generate audio response. Please verify that you have a valid voice set for the character.”)
To help debug this, we need a bit more detail about when this error occurs. Does it happen on a specific type of message or line, or in a particular part of the conversation? Sometimes an unexpected character in the response can cause the TTS to fail and throw this error.
Could you please share:
The Session ID
The Character ID
Whether this always happens in a specific Narrative Design section or trigger
With that information, we can inspect the logs on our side and see why the audio is not being generated.
The error only occurs sometimes, when few devices run the same avatar at the same time. There is a trigger called “Start Trigger” on the narrative design which helps unity (Narrative Design Trigger) to make sure the avatar initialiates the conversation first.
Character id : 0b5d07e0-bb20-11f0-b7b9-42010a7be025
We are still investigating the issue on our side and will share an update as soon as the checks are completed. Thank you for your patience in the meantime.
Hi, We are still facing those issues when multiple ipads (8 devices) run the simulation at the same time. It is high priority for one of our high stake clients, please could you help identify the issue?
I completely understand that this is a high-priority situation for you and your client.
However, dedicated high-priority support is available as part of our Enterprise plans. If you have important clients and projects depending on Convai, I would strongly recommend considering an Enterprise plan so we can better align our support level with your needs.
You can reach our sales team at sales@convai.com to discuss details and find the best option for your use case.
We tested them again today on 9 ipads, it shows the same error. we are not using narrtive design to start the conversation by the avatar. We are using SendTextDataAsync() to do it.
Can please you tell us what has changed and have you addressd the issue?
We are getting no where with this and are losing our client’s confidence. Please could you tell what changes were done when you asked us to try again? On these testing sessions a lot of the times the avatar stopped responding completely mid conversation. Is there a concurrency limit to the voice generation/avatar? All the 9 ipads try to speak to the same avatar with the same character id at the same time during these test sessions and more than 70% of them get these errors or stop responding mid conversation. It’s really an urgent and high priority situation for our relations with the client. We urgently need some direct line of contact and support with the tech team. If we dont get any resolution qucikly will have to choose an alternate option and part our ways from the convai platform.
I completely understand you. The issue you are facing is related to a concurrency limit on the specific voice you are currently using. We have already contacted Google to request an increase to this limit and we are waiting for their update.
As a temporary workaround, I strongly recommend switching this character to a different voice that does not hit the same concurrency limit, and testing again with your setup.
As soon as we have concrete news about the concurrency increase, we will share it with you here.
Thanks a lot for the update. We are testing Azure and Elevenlabs voice now on these avatars. Our Scale subscription plan allow for 15 concurrent session devices at a time.
Could you please let us know which voices to choose so that we get at leaset 15 concurrent voice generation on our avatars?
The current concurrency limitation is only on the Journey voices (the GCP voices that have the word “Journey” in their name). Please avoid using any Journey voices and instead use other available voices.