Turn off automatic bid level correction in the Google Cloud Speech-to-Text API

Question

I am working on a speech recognition task, which includes a “strong” detection of the ability to speak with children, improvement over time ...

I would like to use the Google Cloud Speech to Text API for the ASR detection part. I will then use transcripts of various dimensions to measure progress.

But! Google Speech API sentence level auto-correction sequentially rewrites the previous limb of a pronounced sentence ...

Is there any way to disable autocomplete of this ASR?

I can not get around this problem with the parameters "speechContext", "single_utterance" or "maxAlternatives".

"single_utterance" may work with words, but it corrects spelling errors ..

Any advice in this area?

Answer 1

If you use streaming rather than batch recognition, you should get a response as soon as this part of the audio is transcribed, this does not wait for the rest of the sentence. Then you should just save the first response provided by the stream, and not further corrections.

This means you do not have to wait until isFinal=True.

, , API , , :