test_asr.py

from server import gt
import librosa
ref_audio, _ = librosa.load('/home/salman/salman/minomni_sn21/omega-v2v/miner_models/MiniCPM-o/assets/input_examples/assistant_female_voice.wav', sr=16000, mono=True) # load the reference audio

text = gt(ref_audio, 16_000)
print(text)

# write a code to recursively iterate a directory and subdirectories to transcript all audio .wav files in it
import os
def transcribe_directory():
    for root, dirs, files in os.walk('/home/salman/salman/minomni_sn21/omega-v2v/miner_models/recordings'):
        for file in files:
            if file.endswith('.wav'):
                print(f"Processing file: {file}")
                file_path = os.path.join(root, file)
                audio, sr = librosa.load(file_path, sr=16000, mono=True)
                transcription = gt(audio, sr)
                print(f"Transcription for {file_path}: {transcription}")
                with open(file_path.replace('.wav', '.txt'), 'w') as f:
                    f.write(transcription)


transcribe_directory()