OpenAI transcription vs 3CX Cloud Transcription

[f1_power]

We recently changed from 3CX Cloud Transcription to OpenAI Whisper transcription but we are finding a strange issue where calls when transcribed appear as a slab of text instead of the Caller, time code, user extension etc in the 3CX Cloud transcription. Example attached.

Has anyone seen this before? Is this by design? What are our options here except to go back to 3CX Cloud? Or can on-premise transcription solve this too?

 

[Alejandro_3CX]

Hi @f1_power the difference is due to the stereo mode available with 3CX AI server and Grok which enables the speaker separation and therefore allow identification of who said what with accuracy.