Convert audio files to txt using openai's faster whisper

convert.py

# https://github.com/guillaumekln/faster-whisper
# pip install faster-whisper
# pip install nvidia-cublas-cu11 nvidia-cudnn-cu11
# pip install torch
# Export your library prior to running in WSL or virtual env
# export LD_LIBRARY_PATH=`python3 -c 'import os; import nvidia.cublas.lib; import nvidia.cudnn.lib; print(os.path.dirname(nvidia.cublas.lib.__file__) + ":" + os.path.dirname(nvidia.cudnn.lib.__file__))'`
import os
import torch, gc
import shutil

# get path
# python3 -c 'import os; import nvidia.cublas.lib; import nvidia.cudnn.lib; print(os.path.dirname(nvidia.cublas.lib.__file__) + ":" + os.path.dirname(nvidia.cudnn.lib.__file__))'
os.environ['LD_LIBRARY_PATH'] = "/usr/local/lib/python3.8/dist-packages/nvidia/cublas/lib:/usr/local/lib/python3.8/dist-packages/nvidia/cudnn/lib"

from faster_whisper import WhisperModel

model_size = "large-v2"
#List of valid file extensions
valid_extensions = (".WAV", ".mp3")

# Run on GPU with FP32
model = WhisperModel(model_size, device="cuda", compute_type="float32")
# or run on GPU with INT8, float16, float32
# model = WhisperModel(model_size, device="cuda", compute_type="int8")
# or run on CPU with INT8
# model = WhisperModel(model_size, device="cpu", compute_type="int8")

# Replace 'your_directory' with the actual path to your audio files directory
audio_directory = '/mnt/d/Recordings'
destination_directory = '/mnt/d/Recordings/converted'

for audio_file in os.listdir(audio_directory):
    if audio_file.endswith(valid_extensions):
        source_file = os.path.join(audio_directory, audio_file)
        txt_file = audio_file + ".txt"
        output_file = os.path.join(destination_directory, txt_file)
        segments, info = model.transcribe(source_file, beam_size=5, vad_filter=True)
        # You can process the segments and info here as needed
        print(f"Transcribing {audio_file}...")
        print("Detected language '%s' with probability %f" % (info.language, info.language_probability))

        # Open the file in write mode
        with open(output_file, 'w') as file:
            for segment in segments:
                start_seconds = segment.start
                end_seconds = segment.end

                start_hours = int(start_seconds // 3600)
                start_minutes = int((start_seconds % 3600) // 60)
                start_seconds = start_seconds % 60

                end_hours = int(end_seconds // 3600)
                end_minutes = int((end_seconds % 3600) // 60)
                end_seconds = end_seconds % 60

                # Your code for generating the output
                output = "[%02d:%02d:%02d -> %02d:%02d:%02d] %s" % (start_hours, start_minutes, start_seconds, end_hours, end_minutes, end_seconds, segment.text)
                print(output)

                # Open the file in write mode and save the output
                output += '\n'
                file.write(output)


        print(f"Output saved to {output_file}")
        shutil.move(source_file, destination_directory)
        del segments
        del info
        torch.cuda.empty_cache()

del model