Problem when transcript from microphone

Archives DeepSpeech
#1

Hi all,

I am in windows system and i have tried to use mozzila deep speech to do transcription from microphone.

I have used this line of code to launch it

python mic_vad_streaming.py --model DeepSpeech/test/deepspeech-0.7.4-models.pbmm
–scorer C:/Users/trabe/OneDrive/Documents/DeepSpeech/test/deepspeech-0.7.4-models.scorer

When I said “Hello”: results is
image
When i said 3 times “how are you”
image
When i said 4 times “Hello everyone”:
image

Any help please ?

#2

  1. Take version 0.8 or 0.9 which have better models.

  2. Record audio and then feed it to DeepSpeech to get reproducible results.

  3. If you have an accent, DS might have a hard time recognizing what you are saying. It is unfortunately best for American English.

And no images please, stick to the guidelines.