Call.in Progress.error Assistant Did Not Receive C...
# support
a
what does end reason "Call.in Progress.error Assistant Did Not Receive Customer Audio" mean? We’ve already integrated a soundcheck feature into our app, where the user is prompted to say a specific phrase. This phrase is then transcribed using Deepgram to verify correct pronunciation. However, we still encounter this error when the call starts after the soundcheck.
v
This error occurs when Vapi's assistant cannot detect or process the customer's audio input during a call. Even with a successful soundcheck, this can still happen due to what's known as the "deafness bug" in Vapi. Try this fix:
Copy code
python
# Change your transcriber to an alternative service
assistant = vapi.Assistant(
    transcriber="whisper",  # Instead of "deepgram"
    # Your other configuration...
)
According to [user reports](https://discord.com/channels/1211482211119796234/1343612796717371393), switching from Deepgram to Whisper has resolved this issue in similar cases. Source: - [Deafness bug discussion](https://discord.com/channels/1211482211119796234/1343612796717371393)
This error occurs when Vapi's assistant cannot detect or process the customer's audio input during a call. Even with a successful soundcheck, this can still happen due to what's known as the "deafness bug" in Vapi. Try this fix:
Copy code
python
# Change your transcriber to an alternative service
assistant = vapi.Assistant(
    transcriber="whisper",  # Instead of "deepgram"
    # Your other configuration...
)
According to [user reports](https://discord.com/channels/1211482211119796234/1343612796717371393), switching from Deepgram to Whisper has resolved this issue in similar cases. Source: - [Deafness bug discussion](https://discord.com/channels/1211482211119796234/1343612796717371393)
k
You can try out
a
is this how do you define it?
Copy code
const ephemeralAss = {
...
  startSpeakingPlan: {
    waitSeconds: 1.2,
    smartEndpointingEnabled: true,
  },
  transcriber: {
    provider: "talkscriber",
    model: "whisper",
    language: "en",
  },
};
?
Switching to Whisper has helped a bit—it’s reduced the issue by around 50%—but there are still a few instances where it continues.
k
The error “Call.in Progress.error Assistant Did Not Receive Customer Audio” means no audio was detected from the user during the call, often due to voicemail, stream delays, despite your soundcheck, consider adding a delay before the assistant speaks and improving voicemail detection.. https://docs.vapi.ai/calls/call-ended-reason
a
Could you please clarify what you mean by ‘consider adding a delay before the assistant speaks and improving voicemail detection’? Specifically, how would you suggest we improve voicemail detection? Currently, we’re using Deepgram to prompt the user to say a phrase.
k
To improve voicemail detection, use Vapi’s built-in hybrid detection with configurable parameters, enhance Deepgram accuracy with keyword boosting, and regularly review call logs to fine-tune performance.

https://youtu.be/6TsNK5wce5w?si=QPWyl1LjIXKICIDR

a
the issue still appears to be occurring. from what we’re seeing in our posthog logs, it seems to primarily affect non-Chrome browsers. Does that align with what you’re observing?
k
Send call id
a
befa5bf7-fb8f-4f32-88e7-670065db5a24
h
We're are now also experiencing the same issue for our inbound assistant and Voicemail Detection is OFF. This has only started happening since we switched to Nova 3 General, and adjusted to LiveKit Smart Endpointing and made some adjustments on Idle Timeouts settings. Not sure if these settings have affected this. Call IDs: a641254f-4f65-4ff2-942e-bc592fb5df68 2c05aaf7-cd1c-427c-ab81-fca89d40c3f4 1ffed3da-ee7e-43b6-b322-870d62cff4ca f59f45b1-eb0c-4c1d-9eae-623530e1e87d
k
Thank you for sharing the call id. We will investigate this issue and get back to you shortly
3 Views