116 lines
3.6 KiB
Markdown
116 lines
3.6 KiB
Markdown
# Audiobook Creator
|
|
|
|
AI-powered audiobook generator using the [Kokoro TTS](https://github.com/hexgrad/kokoro) model.
|
|
Generates high-quality narrated `.wav` files from plain-text novels, with a GUI tool for auditing and fixing proper noun pronunciations per book.
|
|
|
|
---
|
|
|
|
## Features
|
|
|
|
- **Multi-book support** — each book's proper nouns, fixes, and audio are fully isolated
|
|
- **Proper Noun GUI** — hear every extracted name, mark it correct or type a phonetic fix
|
|
- **Audiobook generation** — one `.wav` per chapter, GPU-accelerated via CUDA
|
|
- **In-GUI extraction** — click one button to run NLP extraction and generate audio, no separate scripts needed
|
|
- **Apply Fixes** — writes a TTS-ready copy of the source text with all phonetic substitutions applied
|
|
|
|
---
|
|
|
|
## Project structure
|
|
|
|
```
|
|
Audio Text for Novel Lightbringer/ ← multi-file book (chapters as .txt)
|
|
Audio Master Nem Full.txt ← single-file book
|
|
|
|
gui_proper_noun_player.py ← proper noun auditing GUI
|
|
create_audiobook_lightbringer.py ← generate Lightbringer audiobook chapters
|
|
create_audiobook_nem.py ← generate Nem audiobook chapters
|
|
|
|
output_audiobook_lightbringer/ ← chapter WAV output
|
|
output_audiobook/ ← Nem WAV output
|
|
output_proper_nouns/<book>/ ← manifest + JSON fix data per book
|
|
proper_nouns_audio/<book>/ ← word audio + replacements cache per book
|
|
|
|
requirements.txt
|
|
setup_windows.bat ← one-click Windows setup
|
|
run_gui.bat ← launch GUI on Windows
|
|
run_audiobook.bat ← generate audiobook on Windows
|
|
```
|
|
|
|
---
|
|
|
|
## Setup (Linux / Mac)
|
|
|
|
```bash
|
|
python3.11 -m venv .venv
|
|
source .venv/bin/activate
|
|
pip install torch --index-url https://download.pytorch.org/whl/cu124 # CUDA 12.4
|
|
pip install -r requirements.txt
|
|
python -m spacy download en_core_web_sm
|
|
```
|
|
|
|
> For CPU-only: replace the torch line with `pip install torch`
|
|
|
|
---
|
|
|
|
## Setup (Windows)
|
|
|
|
See [SETUP_WINDOWS.md](SETUP_WINDOWS.md) for a step-by-step guide aimed at non-technical users.
|
|
|
|
---
|
|
|
|
## Usage
|
|
|
|
### Proper Noun GUI
|
|
|
|
```bash
|
|
.venv/bin/python gui_proper_noun_player.py
|
|
```
|
|
|
|
1. Select a book from the dropdown
|
|
2. Click **⚙ Extract & Generate Audio** — extracts proper nouns via spaCy and generates a TTS clip for each one
|
|
3. Click words in the Review list to hear them; press Enter to mark correct or type a phonetic replacement first
|
|
4. Click **⇄ Apply Fixes to Text** to write a pronunciation-corrected copy of the source file
|
|
|
|
### Generate Audiobook
|
|
|
|
```bash
|
|
# All chapters
|
|
.venv/bin/python create_audiobook_lightbringer.py
|
|
|
|
# List chapters only
|
|
.venv/bin/python create_audiobook_lightbringer.py --list
|
|
|
|
# Preview clips
|
|
.venv/bin/python create_audiobook_lightbringer.py --preview
|
|
|
|
# Specific chapters
|
|
.venv/bin/python create_audiobook_lightbringer.py 0 1 2
|
|
```
|
|
|
|
---
|
|
|
|
## Dependencies
|
|
|
|
| Package | Purpose |
|
|
|---|---|
|
|
| `kokoro` | Kokoro-82M TTS model |
|
|
| `torch` | GPU inference |
|
|
| `soundfile` / `sounddevice` | Audio I/O |
|
|
| `numpy` | Audio array operations |
|
|
| `spacy` + `en_core_web_sm` | Proper noun extraction (NER + PROPN) |
|
|
| `wordfreq` | Common-word filter during extraction |
|
|
|
|
---
|
|
|
|
## Output
|
|
|
|
| Path | Contents |
|
|
|---|---|
|
|
| `output_audiobook_lightbringer/` | `chapter_01_homecoming.wav`, … |
|
|
| `output_proper_nouns/<book>/manifest.json` | Word → WAV filename map |
|
|
| `output_proper_nouns/<book>/pronunciation_fixes.json` | `{"Nephi": "Kneephi", …}` |
|
|
| `output_proper_nouns/<book>/correct_words.json` | Words confirmed correct |
|
|
| `proper_nouns_audio/<book>/` | Per-word audio clips |
|
|
| `proper_nouns_audio/<book>/replacements_cache/` | Cached phonetic fix clips |
|
|
|