Chapterize Audio

This script generates chapters for an audio file by detecting silence. The chapters are stored to a CUE file and a JSON file.

Prerequisites

• Python
• FFmpeg (Make sure it is installed and added to the system's PATH)

Installation

pip install -r requirements.txt

Usage

Example usage:

python chapterize.py -i <audio.mp3>  -t <dB threshold> -d <silence duration>

Options:

options:
  -h, --help            show this help message and exit
  -i INPUT_AUDIO, --input_audio INPUT_AUDIO
                        The input audio file
  -t THRESHOLD, --threshold THRESHOLD
                        The silence threshold in dB (default: -30)
  -d DURATION, --duration DURATION
                        The minimum silence duration in seconds (default: 2.5)
  -l LABEL, --label LABEL
                        The label to prefix the chapter nr. (default: Part)
  --transcribe, --no-transcribe
                        Enable transcribing of chapter entry (default: True)
  --language LANGUAGE   The language to use for transcribing (default: english)
  --transcript-duration TRANSCRIPT_DURATION
                        The duration of the audio to transcribe (default: 3)

Console Output:

$ python chapterize.py -i audio.mp3 -t -30 -d 2.5
Part duration: <chapter duration> Next part start: <chapter start time>, <transcribed chapter start>
...

File Output:

./audio.cue
./chapters.json

CUE file output:

FILE "audio.mp3" MP3
TRACK 1 AUDIO
    TITLE "Part 1"
    INDEX 01 00:00:00
TRACK 3 AUDIO
    TITLE "Part 3"
    INDEX 01 00:14:37
TRACK 4 AUDIO
    TITLE "Part 4"
    INDEX 01 01:56:36

JSON file output:

[
    {
        "id": "0",
        "title": "Part 1",
        "start": "00:00:00",
        "end": "00:14:37"
    },
    {
        "id": "1",
        "title": "Part 3",
        "start": "00:14:37",
        "end": "01:56:36"
    },
    {
        "id": "2",
        "title": "Part 4",
        "start": "01:56:36",
        "end": "02:00:00"
    }
]

Todo list

• Option to fix timing of existing chapters
• Option to overwrite chapters in audiobookshelfs metadata.json directly

Acknowledgements

• FFmpeg