Awaaz v1.0.0
Initial Release
Awaaz is a resumable, API-first audiobook generation service. Upload EPUBs or plain text, pick a voice, and generate a stitched MP3 audiobook in chunks.
What's Included
- FastAPI backend with OpenAI-compatible
/v1/audio/speechendpoint - Web UI for document library, voice settings, job monitoring, and downloads
- Custom QueueWorker for round-robin, resumable chunk processing
- Support for Supertonic and Kokoro TTS backends, plus a custom backend option
- Docker Compose setup for API, worker, frontend, and TTS services
Highlights in this release
- Live chunk-count preview in the generation form
- Semantic chunking modes (paragraph, line, sentence) with no implicit packing
- Character-limit mode for dense packing
- Progress tracking with pause/resume/cancel
- Cover image extraction from EPUBs
- Collection-based document organization
Development Quality
- 34 Python tests passing
- 4 frontend tests passing
- Strict mypy clean across 21 source files
- Ruff linting clean