wav2vec-label.py

import audeer
import audonnx
import librosa
import glob
import os

url = 'https://zenodo.org/record/6221127/files/w2v2-L-robust-12.6bc4a7fd-1.1.0.zip'
model_path = './model.zip'

audeer.download_url(
    url, 
    model_path, 
    verbose=True,
)

audeer.extract_archive(
    model_path, 
    ".", 
    verbose=True,
)

model = audonnx.load(".")

SAMPLING_RATE = 16000

files = glob.glob("output/*.wav")
target_folder = os.path.join(".", "kit-wav2vec2-label-16k")

for i, path in enumerate(files):
    wav, _ = librosa.load(path, sr=SAMPLING_RATE, mono=True)
    source = model(wav, SAMPLING_RATE)
    new_name = ("%.3f-%.3f-%.3f__" % (source['logits'][0][0], source['logits'][0][1], source['logits'][0][2])) + os.path.basename(path)
    new_path = os.path.join(target_folder, new_name)
    
    os.rename(path, new_path)