Datasets:
How to convert to normal audio files?
#2
by
Mavisfsew
- opened
Hi, Please i downloaded this dataset and I want to extract the files from the parquet but the instructions provided and the stracture of the data is not the same.
Here is the script
import pandas as pd
import os
import csv
import numpy as np
Read the Parquet file into a pandas DataFrame
df = pd.read_parquet('train-00000-of-00001-49c48937e7376123.parquet')
Create folders to store the audio files and metadata
audio_folder = 'Wavs'
os.makedirs(audio_folder, exist_ok=True)
Open the metadata.csv file for writing
metadata_file = 'metadata.csv'
with open(metadata_file, 'w', newline='') as csvfile:
fieldnames = ['line_id', 'text', 'speaker_id']
writer = csv.DictWriter(csvfile, fieldnames=fieldnames)
writer.writeheader()
# Iterate over each row in the DataFrame
for idx, row in df.iterrows():
# Get the audio array and sampling rate
audio_array = row['audio']['bytes']
sampling_rate = 48000
# Convert the audio array to bytes
audio_bytes = np.array(audio_array, dtype=np.float32).tobytes()
# Save the audio file
audio_path = os.path.join(audio_folder, f"{row['line_id']}.wav")
with open(audio_path, 'wb') as f:
f.write(audio_bytes)
# Write metadata to the CSV file
writer.writerow({
'line_id': row['line_id'],
'text': row['text'],
'speaker_id': row['speaker_id']
})