you can use `librosa`, the task is to split array on chunks which size exactly equals sample rate multimplies 5 seconds
import librosa
import soundfile as sf
audio_file = "example.mp3"
y, sr = librosa.load(audio_file, sr=None)
# chunk duration 5 seconds
chunk_duration = 5
chunk_samples = int(chunk_duration * sr)
chunks = [y[i:i + chunk_samples] for i in range(0, len(y), chunk_samples)]
for i, chunk in enumerate(chunks):
output_file = f"chunk_{i}.mp3"
sf.write(output_file, chunk, sr)
you can split array in chunks just using index spliting, in order to get 2 seconds you need 2 * sample rate values from array and then step ahead
import librosa
import soundfile as sf
audio_file = "0.wav"
y, sr = librosa.load(audio_file, sr=None)
# chunk duration 2 seconds
chunk_duration = 2
chunk_samples = int(chunk_duration * sr)
chunks = [y[i:i + chunk_samples] for i in range(0, len(y), chunk_samples)]
for i, chunk in enumerate(chunks):
output_file = f"chunk_{i}.wav"
sf.write(output_file, chunk, sr)
