I downloaded the dataset and try to merge all audio, because I want to try split with different number, however I only 34k audio samples, which is not even 50% of 200k

audio path

dataset_path = '/content/CommonLanguage/common_voice_kpd'

Function to load audio data

def load_audio(file_path):
try:
audio_data, sample_rate = librosa.load(file_path, sr=None)
return audio_data, sample_rate
except Exception as e:
print(f"Error processing file {file_path}: {e}")
return None, None

Function to load audio data with multithreading

def load_audio_multithreaded(dataset_path):
audio_list = []
labels = []

def process_subfolder(subfolder_path, label):
    if os.path.isdir(subfolder_path):
        for file in os.listdir(subfolder_path):
            file_path = os.path.join(subfolder_path, file)
            if file_path.endswith(".wav"):
                futures.append(executor.submit(load_audio, file_path))
                labels.append(label)

def process_folder(folder_path, label):
    if os.path.isdir(folder_path):
        for subfolder in os.listdir(folder_path):
            subfolder_path = os.path.join(folder_path, subfolder)
            if os.path.isdir(subfolder_path):
                process_subfolder(subfolder_path, label)

with ThreadPoolExecutor(max_workers=4) as executor:
    futures = []
    start_time = time.time()
    for language_folder in os.listdir(dataset_path):
        language_folder_path = os.path.join(dataset_path, language_folder)
        if os.path.isdir(language_folder_path):
            train_path = os.path.join(language_folder_path, 'train')
            test_path = os.path.join(language_folder_path, 'test')
            dev_path = os.path.join(language_folder_path, 'dev')

            # Process files in train, test, and dev folders
            process_folder(train_path, language_folder)
            process_folder(test_path, language_folder)
            process_folder(dev_path, language_folder)

    # Retrieve results from futures
    for future in futures:
        audio_data, sample_rate = future.result()
        if audio_data is not None:
            audio_list.append((audio_data, sample_rate))

    end_time = time.time()
    print(f"Total time taken to load audio data: {end_time - start_time:.2f} seconds")

return audio_list, labels

Load audio data

audio_list, labels = load_audio_multithreaded(dataset_path)