WAV Processing

Audio effects

python3 src/audio_effects.py -i audios/eng.wav -s 0.3 -v 2.5 -o audios/out.wav

• -i/--input - path to input audio.wav
• -s/--speed - speed factor > 0.0. Factor by which to adjust speed of input. Values greater than 1.0 compress waveform in time, whereas values less than 1.0 stretch waveform in time
• -v/--volume - volume factor > 0.0. Factor by which to adjust volume of input. Values greater than 1.0 make audio louder, whereas values less than 1.0 make audio quieter
• -o/--output - path where to save output audio.wav

Speech to text

python3 src/audio_effects.py -i audios/eng.wav -m ./models/ -o transcipts.json -l ru

• -i/--input - path to input audio.wav. Only rus or eng
• -m/--models-dir - path to models dir. See descriptions below
• -d/--device - where to do computation. cpu or cuda:{x}
• -l/--lang - predefined language ru or en. If not stated, language recognition will be performed
• -o/--output - path where to save JSON with transcipts. Optionally

Models dir

Models dir must have following structure

- multilang_stt.pt
- voxligua:
    - classigier.ckpt
    - config.json
    - embedding_model.ckpt
    - hyperparams.yaml
    - label_encoder.ckpt
    - label_encoder.txt
    - langs.json
    - normalizer.ckpt

models can be downloaded from here