Notifications
Clear all
Topic starter
I'm attempting to construct a Customer Service Assistant that connects to a database, utilizing Twilio, ElevenLabs, Supabase, and GPT-4o. However, I'm encountering issues getting the workflow to function correctly.
Here's a snapshot of where the execution is currently getting stuck:
Whenever I dial my Twilio number, the workflow triggers, but no audio is transmitted back to the call, nor is it sent to ElevenLabs. I've spent considerable time seeking solutions from AI assistants, but without success. Each suggestion has led to a new problem.
I've verified that all ports are connected, that wsproxy is configured correctly, and that the calls are establishing via the WS. I'm uncertain how to proceed.
It appears to be getting stuck on this node, failing to pass information through.
The audio consistently goes into the 'false branch,' and I'm unsure why. The WebSocket is connected, but the calls are silent. I cannot hear anything, and it seems the workflow isn't receiving the audio either.
It appears the audio is not being processed.
Any assistance would be greatly appreciated, as this is causing significant frustration.
Thank you in advance for any help in resolving this.
Best regards,
JSY
Posted : 30/06/2025 12:49 pm
Topic starter
I've now resolved the previous issue and encountered a new one.
When I transmit the audio/mulaw file using model_id: ‘scribe_v1’, the callin.io API returns a 400 error with the message: "File audio.mulaw is corrupted. Please ensure it is playable audio."
If I attempt to use the eleven_multilingual_v2 model, which is more appropriate for telephony audio, callin.io responds with a 400 error stating: "'elevenmultilingualv2' is not a valid modelid. Only 'scribev1' and 'scribev1experimental' are currently available."
So, the problem now seems to be with the audio file format, and I'm again uncertain about how to proceed.
Posted : 30/06/2025 1:07 pm