Search code examples
amazon-web-servicesamazon-transcribe

Amazon Transcribe - Should I use audio pre-processing or does Amazon do that?


I'm currently experimenting with the Amazon Transcribe service and I'm wondering if I should pre-process the audio files before sending them to Amazon Transcribe. I couldn't find anything in the documentation so I would assume they don't apply any pre-processing.

For example I could apply a lowpass and a highpass filter to enhance voices and cut out unwanted noises. I'm wondering if I have to do this or if Amazon already does this for me?

I already experimented a bit and I couldn't see a real improvement if I applied the filters but that could also mean that my audio was already good enough... I can't do extensive experimenting because it costs me money everytime I do that...

It would be nice if someone could share their experience with this service.


Solution

  • According to Amazon Transcribe FAQs - Yes, the quality and content of the audio signal (including but not limited to factors such as background noise, overlapping speakers, accented speech, or switches between languages within a single audio file) may affect the accuracy of service output.

    You can find more detail info here.