0

Is it possible to give a text sentence and a sound sample to Mozilla-DeepSpeech software and get a matching percentage of both ?

It to make a quizz software. Program know the response and ask user to pronounce it in the microphone.

FabienM
  • 3,421
  • 23
  • 45

1 Answers1

1

No, DeepSpeech will simply generate the best it can "hear" from the audio you provide. You could use it's test functionality to do that, but that is a bit of an overkill.

Try a tool like BenchmarkSTT for that.

Olaf
  • 158
  • 7
  • Ok, my objective is not to make a benchmark, just a simple quizz software that ask for spoken response. Then, the software know which response is correct. – FabienM Nov 04 '20 at 14:41
  • 1
    I think I get it. Please ask that on Discourse as it is easier to point to other answers as you'll need to tweak the system a bit: https://discourse.mozilla.org/c/mozilla-voice-stt/247 – Olaf Nov 04 '20 at 15:42