Awaaz - आवाज़
A self-hosted app with beautiful UI for audiobook creation.
12 followers
A self-hosted app with beautiful UI for audiobook creation.
12 followers
Awaaz turns EPUBs, TXT files, or pasted text into MP3 audiobooks without the usual TTS timeouts or progress loss. Progress is persisted, failed chunks are retried. Built for readers, creators, and developers who want a self-hosted, library-style audiobook pipeline. Key features • Import anything: EPUB/TXT uploads or pasted text. • Fine-tune narration: choose backend, model, voice, speed etc. • Durable jobs. • Multiple interfaces. • Pluggable backends. • Self-hosted.
Free
Launch Team
Render: Get up to $100K in startup creditsThe fastest way to launch and scale your startup in the cloud
Promoted
Hey everyone 🙋🏻♂️, I'm Sharad.
I've open-sourced the repo for Awaaz ... a resumable, self-hosted, API-first audiobook generator for long-form content. The UI is inspired by ElevenReader, which I use most of the time apart from local models.
I'll be adding advanced features over the next few days. I currently use these features through a PoC I made for my own use case, and I now need to migrate them to Awaaz.
What it does: Awaaz turns EPUB, TXT, or pasted text into MP3 audiobooks. Long documents become durable synthesis jobs instead of one fragile TTS request. Text is chunked, progress is persisted, failed work can be retried, and completed WAV chunks survive worker restarts.
It's API-first so you can drive it entirely from curl.
Stars, issues, and feedback welcome! 🙂
Self-hosted audiobook conversion is something I keep meaning to set up, and the chunk retry plus persisted progress sounds like it actually solves the half-finished TTS job problem. Curious how the EPUB parsing holds up on files with weird formatting.
@laydaikgzygty EPUB parsing is handled via calibre cli so we do not have to do anything with it, its the best way there is.
How does it handle really long EPUBs - does the chunking keep chapter boundaries intact or could sentences get split mid-thought across audio segments?
@askar_gamz10917 It keeps them intact, there are ways to configure the splitting if you really want to do.
We have
Para
Sentence
Line
Character - requires you to configure character limit so if a sentence exceeds that limit safe word cound less then specified will be a single chunk
will be adding more if requested.
Finally gave this a spin with a hefty EPUB and the chunk retry actually held up where my old TTS scripts always choked halfway. Picking voice and speed per job feels right for long-form stuff.
@sezerdasekwcwb Amazing! Thanks for the review. Will be adding your review to the landing page if you don't mind. If you need some improvements let me know.
love that the progress is actually persisted between sessions, that is the kind of detail most tts tools completely ignore and it makes all the difference when you are converting a 500 page book. solid execution here.
@aykut597715 Thanks a lot for your feedback. I frequently ran into those issues myself and it was a headache. I'm actively maintaining this so let me know if you wanna suggest some feature or have any feedback.