Amplify Speech to Text

Question

Hi All,

I am trying to use Amplify Prediction transcription option to convert a WAV file to text. I have a File Object and I converted it to ArrayBuffer and passed it to the below code as source.

```
Predictions.convert({
          transcription: {
            source: {
              buffer,
            },
            language: "en-US",
          }
        })
```
But it says invalid source. I tried to use AudioBuffer as well and getting same error. I even used Int8Array and other options but still getting invalid source error. Please let me know how to pass the File object to transcribe?

Didier_Durand · Answer

Hi.

Text to speech has to be done with Amazon Transcribe not Amplify: https://aws.amazon.com/pm/transcribe/

Best,

Didier

Antonio_Lagrotteria · Answer

Check out my blog which leverage Amplify predictions (Polly) and GitHub repo.

https://levelup.gitconnected.com/build-your-personal-speaker-assistant-with-amplify-and-chatgpt-8b6433fea042

I think the source fir prediction could be as simple as a plain string

Subalalitha · Answer

I am using Amplify for some other purpose, but Amplify also allows Transcribe via Predictions
https://docs.amplify.aws/react/build-a-backend/more-features/predictions/transcribe-audio/
Just wanted to know how to do this in the above service

Subalalitha · Answer

Thanks for the link. But your code uses text to speech. I need speech to text. In the doc they said the input should be PCM Audio byte buffer. But it is not accepting that or any other type.

Amplify Speech to Text

Relevant content