To test the updated denoise_and_vad_audio.py locally.
Note: This script is tested on an .tsv file.
.tsv file contains filename/path for the audio.
This PR deals with additional arguments for SeperateSpeech i.e. --config : configuration.yaml file and --pth-model : path for the model .pth file.
How to Test:
Locally, just run the following:
To test master64:
python denoise_and_vad_audio.py --audio-manifest $INPUT_MANIFEST --output-dir $OUTPUT_DIR --denoise --vad
($INPUT_MANIFEST is path for .tsv file)
To test SeperateSpeech:
python denoise_and_vad_audio.py --audio-manifest $INPUT_MANIFEST --output-dir $OUTPUT_DIR --model SeperateSpeech --config /path/to/config.yaml --pth-model /path/to/model.pth --denoise --vad
It will generate two output audios: denoise and vad audio files in their respective directory.
lpw0
commented
7 months ago
Before submitting
What does this PR do?
Integrates TFGridNet from ESPNET into fairseq2.
Testing
To test the updated denoise_and_vad_audio.py locally.
Note: This script is tested on an .tsv file. .tsv file contains filename/path for the audio.
This PR deals with additional arguments for SeperateSpeech i.e. --config : configuration.yaml file and --pth-model : path for the model .pth file. How to Test: Locally, just run the following:
To test master64: python denoise_and_vad_audio.py --audio-manifest $INPUT_MANIFEST --output-dir $OUTPUT_DIR --denoise --vad
($INPUT_MANIFEST is path for .tsv file)
To test SeperateSpeech: python denoise_and_vad_audio.py --audio-manifest $INPUT_MANIFEST --output-dir $OUTPUT_DIR --model SeperateSpeech --config /path/to/config.yaml --pth-model /path/to/model.pth --denoise --vad
It will generate two output audios: denoise and vad audio files in their respective directory.