chunk_audio.py

# Adapted from https://github.com/timothymeehan/ambient_music_generator/blob/master/code/chunk_audio_overlap.py

import os
import sys
import argparse
from pydub import AudioSegment
from pydub import silence

parser = argparse.ArgumentParser(description='Split a .wav file into chunks')
parser.add_argument('--input_file', type=str, required=True, help='Path to the input .wav file')
parser.add_argument('--output_dir', type=str, required=True, help='Output directory for the chunks')
parser.add_argument('--chunk_length', type=int, default=8000, help='Output chunk size in milliseconds')
parser.add_argument('--overlap', type=int, default=0, help='Overlap between consecutive chunks in milliseconds')

args = parser.parse_args()

input_file = args.input_file
output_dir = args.output_dir
chunk_length = args.chunk_length
overlap = args.overlap

# Create output dir if it doesn't exist
if not os.path.exists(output_dir):
    os.makedirs(output_dir)

# Checks for silent chunks
def is_silent(chunk, chunk_length):
    # if at least half of chunk is silence, mark as silent
    silences = silence.detect_silence(
		chunk,
		min_silence_len=int(chunk_length/2),
		silence_thresh=-64)
    if silences:
        return True
    else:
        return False

# Load file and set directories
audio = AudioSegment.from_wav(input_file)
input_filename = input_file.split('/')[-1].replace('.wav', '')

# Get length of audio
audio_len = len(audio)

# Initialize start and end seconds to 0
start = 0
end = 0

# Iterate from 0 to end of the file,
# with increment = chunk_length
cnt = 0
num_silent = 0
flag = 0 # Break loop once we reach the end of the audio
for i in range(0, 8 * audio_len, chunk_length):

	# At first, start is 0, end is the chunk_length
	# Else, start=prvs end-overlap, end=start+chunk_length
	if i == 0:
		start = 0
		end = chunk_length
	else:
		start = end - overlap
		end = start + chunk_length

	# Set flag to 1 if endtime exceeds length of file
	if end >= audio_len:
		end = audio_len
		flag = 1

	# Storing audio file from the defined start to end
	chunk = audio[start:end]
	if flag == 0:
		cnt = cnt + 1
		chunk_is_silent = is_silent(chunk, chunk_length)
		if chunk_is_silent:
			print('Chunk {} is silent, omitting it.'.format(cnt))
			num_silent = num_silent + 1
		else:
			filename = input_filename + f'_chunk_{cnt}.wav'
			chunk.export(os.path.join(output_dir, filename), format="wav")
			print("Processing chunk " + str(cnt) + ". Start = "
					+ str(start) + " end = " + str(end))

print('\n')
print("Finished chunking {}.".format(input_file))
print("{} chunks processed, {} were silent.".format(cnt, num_silent))
print("Saved {} chunks to {}.".format(cnt - num_silent, output_dir))