brew install ffmpeg

sudo apt-get install ffmpeg

pip install pydub

# mac: brew install ffmpeg
# linux: sudo apt-get install ffmpeg
# pip install pydub
from pydub import AudioSegment

def detect_leading_silence(sound, silence_threshold=-50.0, chunk_size=10):
    '''
    sound is a pydub.AudioSegment
    silence_threshold in dB
    chunk_size in ms

    iterate over chunks until you find the first one with sound
    '''
    trim_ms = 0 # ms

    assert chunk_size > 0 # to avoid infinite loop
    while sound[trim_ms:trim_ms+chunk_size].dBFS < silence_threshold and trim_ms < len(sound):
        trim_ms += chunk_size

    return trim_ms

# 16 KB
sound = AudioSegment.from_file("example.mp3", format="mp3")

start_trim = detect_leading_silence(sound, silence_threshold=-40.0)
end_trim = detect_leading_silence(sound.reverse(), silence_threshold=-40.0)
duration = len(sound)
trimmed_sound = sound[start_trim:duration-end_trim]

trimmed_sound.export("example2.mp3", format="mp3")