mirror of
https://github.com/storytold/storyteller-ml.git
synced 2026-10-13 23:53:02 +00:00
michael chung
40 lines
1.7 KiB
Python
40 lines
1.7 KiB
Python
import os
|
|
import torchaudio
|
|
|
|
def load_and_resample_audio(file_path, target_sample_rate=24000):
|
|
# Load audio file
|
|
waveform, sample_rate = torchaudio.load(file_path)
|
|
|
|
# Resample to target_sample_rate if necessary
|
|
if sample_rate != target_sample_rate:
|
|
resampler = torchaudio.transforms.Resample(orig_freq=sample_rate, new_freq=target_sample_rate)
|
|
waveform = resampler(waveform)
|
|
|
|
# Convert to mono by averaging the channels if it's not already mono
|
|
if waveform.shape[0] > 1:
|
|
waveform = waveform.mean(dim=0, keepdim=True)
|
|
|
|
return waveform, target_sample_rate
|
|
|
|
def process_and_overwrite_audio_files(folder_paths, target_sample_rate=24000):
|
|
for folder_path in folder_paths:
|
|
# Iterate through each file in the folder
|
|
for filename in os.listdir(folder_path):
|
|
file_path = os.path.join(folder_path, filename)
|
|
|
|
# Check if the path is an actual file to prevent trying to load directories
|
|
if os.path.isfile(file_path):
|
|
try:
|
|
# Load, resample, and convert to mono
|
|
waveform, _ = load_and_resample_audio(file_path, target_sample_rate)
|
|
|
|
# Overwrite the original file with the processed audio
|
|
torchaudio.save(file_path, waveform, target_sample_rate)
|
|
print(f"Processed and overwritten: {file_path}")
|
|
except Exception as e:
|
|
print(f"Error processing {file_path}: {e}")
|
|
|
|
# Example usage:
|
|
# Replace these with the paths to your folders containing audio files
|
|
folder_paths = ['ood', 'wavs','val']
|
|
process_and_overwrite_audio_files(folder_paths) |