build_timecodes_based_on_silence

build_timecodes_based_on_silence.py

from pydub import AudioSegment, silence
import pandas as pd

def build_segments(audio, length_segment=10, dbfs=0):
    silences = silence.detect_silence(audio, min_silence_len=1000, silence_thresh=dbfs-16)
    dfp_silences = pd.DataFrame(silences, columns = ["start_timecode", "end_timecode"])

    threshold_segment = int(length_segment * 60 * 1000)
    first_timecode = 0
    last_timecode = int(audio.duration_seconds * 1000)

    buffer = []
    for idx, row in dfp_silences.iterrows():

        delta_time_segment = row["start_timecode"] - first_timecode
        if delta_time_segment >= threshold_segment:
            buffer.append([first_timecode, row["start_timecode"]])
            first_timecode = row["end_timecode"]
            
    #Check the last segments
    dfp_new_segments = pd.DataFrame(buffer, columns = ["start_timecode", "end_timecode"])
    if dfp_new_segments.at[len(dfp_new_segments)-1, "end_timecode"] != last_timecode:
        dfp_new_segments.at[len(dfp_new_segments)-1, "end_timecode"] = last_timecode
    return dfp_new_segments
  
audio = AudioSegment.from_mp3(f"location_of_file.mp3")
dfp_timecodes = build_segments(audio, length_segment=10)