This is about performing an interesting demo of open source speech recognition on the command line. Deepspeech was being actively developed by Mozilla for a few years.
Install portaudio-devel
dnf install portaudio-devel python3-devel python3.8
Create Virtual Environment
python3.8 -m venv <custom>
Activate virtual environment
source <custom>/bin/activate
Install python requirements
pip install -U deepspeech deepspeech-tflite pyaudio webrtcvad halo numpy scipy pyautogui
Download the mic vad streaming program
https://github.com/mozilla/DeepSpeech-examples/blob/r0.9/mic_vad_streaming/mic_vad_streaming.py
Download the tflite file
https://github.com/mozilla/DeepSpeech/releases
Run mic_vad_streaming.py
In the directory of these downloaded files, run the command:
python mic_vad_streaming.py --model ./deepspeech-0.9.x-models.tflite --scorer ./deepspeech-0.9.x-models.scorer
Now speak into your microphone to see how accurate the recognition is.
No comments:
Post a Comment