DANIEL CORSANO

Installation

Install via pip (Python 3.10-3.13 required):

pip install audiobook-reader

The package is audiobook-reader, the command it installs is reader.

PyPI Package: https://pypi.org/project/audiobook-reader/

GitHub Repository: https://github.com/danielcorsano/reader

Requirements

• Python 3.10-3.13
• FFmpeg (brew install ffmpeg on macOS, winget install ffmpeg on Windows, sudo apt install ffmpeg on Linux)
• Kokoro voice models (~310MB, downloads automatically on first use to system cache)
• Supports macOS, Windows, Linux

Features

• 54 voices in 9 languages (American English, British English, Japanese, Mandarin Chinese, Spanish, French, Hindi, Italian, Brazilian Portuguese)
• Voice blending: Mix multiple voices for unique characteristics
• Format support: EPUB, PDF, TXT, Markdown, RST input → MP3, WAV, M4A, M4B output
• Performance optimization: Hardware detection for NPU/GPU/CPU acceleration (approximately 10x real-time on Apple Silicon with Neural Engine)
• Batch processing: Glob patterns for converting multiple files
• Scriptable: Integrate into automation workflows
• Checkpoint resumption: Resume interrupted conversions
• Chapter detection: Automatically identifies chapters in input files
• Progress tracking: Real-time conversion status in terminal
• NEW FEATURE: Text stripping: use the strip command to automatically detect and remove non-content like bibliography, table of contents etc. Then use the automatic content parser to choose which sections of the text you want to have read. Available only in CLI for now.

Usage Examples

Basic conversion:

reader convert --file book.epub

Interactive text cleaning (notes, table of contents, ISBN number etc.) followed by optional conversion:

reader strip mybook.epub

The strip command detects chapters (marked EPUB structure → headings → formatting fallback), then a 5-signal classifier (title keywords, EPUB metadata, content patterns, prose density, relative length) suggests which front and back matter to remove — copyright pages, table of contents, bibliography, index, acknowledgments, about the author.

You then tune the boundaries, front first and back second:

[1] strip more  [2] strip less  [3] OK  [4] skip auto-strip (go to manual)

The back menu adds [5] Back to redo the beginning, and the ending is only previewed if you ask for it, so nothing gets spoiled. Optional manual refinement uses strip/keep syntax:

s 0, 6-8   → Strip chapters 0, 6, 7, 8 (keep the rest)
k 1-5      → Keep chapters 1-5 only (strip the rest)

The cleaned text is saved as mybook_stripped.txt next to the original (output is always plain text — EPUB and PDF are extracted), and you are then offered conversion, with an interactive dialog for language, voice, and speed.

Specify voice:

reader convert --file book.pdf --voice af_sky

Choose where the audiobook lands (downloads is the default, same puts it next to the source):

reader convert --file book.epub --output-dir same

Batch conversion:

reader batch add books/*.epub --output-dir output_dir/
reader batch process --max-workers 4

Custom speed and format:

reader convert --file novel.txt --speed 1.2 --format m4b --voice am_adam

List available voices:

reader voices

Advanced Usage

Configuration file support: Defaults are saved permanently with the config command (run it with no options to print the current settings):

reader config --voice bf_emma --speed 1.1 --format m4b

These live in ~/.config/audiobook-reader/config.yaml. For per-project defaults, drop a .reader.yaml next to your books — it is picked up automatically by searching upward from the file being converted:

tts:
  voice: bf_emma
  speed: 1.1
audio:
  format: m4b
output_dir: ~/audiobooks/

Character voices (beta): Assign different voices to characters in dialogue. See documentation for syntax.