Does the Fast Transcription service (part of speech to text) have possible long delays before processing like batch processing does?

Neil Payne 0

Hello

Im trying to decide which service i should use to convert speech to text

im currently using prerecorded audio clips from a mobile app being sent to the API and then needs to send the text result back as soon as possible.

Real-time Transcription is good for non prerecorded audio, it converts in real time so for a 30 second prerecorded audio file it takes another 30 seconds to transcribe (even if using continuous recognition where i get the data back in streams, it still takes 30 seconds to complete.) Edit: it seems to take half the time, so 15 seconds for 30 sec audio clip

Batch Transcription. not sure if this is quicker to convert? The documentation states there could be 30min delay is starting the batch job in peak hours, so rules that out

Fast Transcription. this may be the way to go, there are no warnings about delays. how fast is this process?

Thank you!

SriLakshmi C (Quadrant Resource LLC) 170 Reputation points Microsoft Vendor

2024-10-03T07:05:13.7766667+00:00

Hello Neil Payne,

Welcome to Microsoft Q&A forum!

As I understood that among these servces Real-time Transcription, Batch Transcription, Fast Transcription you're trying to decide which service to use for converting prerecorded audio clips to text. You need the text results as quickly as possible.

As you mentioned In Batch Transcription there could be 30min delay is starting the batch job in peak hours, so rules that out which might not be ideal for your needs.

Fast Transcription service includes our latest end-to-end model technologies, with best quality and super high Speed Factor leveraging GPU inference (typically it can transcribe a 20-minute audio file in less than 1 minute).

For more information please refer the link Use the fast transcription API.

Please let us know if you have any further questions. Thank you!

Share via

Does the Fast Transcription service (part of speech to text) have possible long delays before processing like batch processing does?

Your answer