Launching today
Audar-TTS-V1
Open-weight Arabic-first expressive text-to-speech
5 followers
Open-weight Arabic-first expressive text-to-speech
5 followers
Audar-TTS-V1 is an Arabic-first, open-weight expressive text-to-speech model. Control delivery with inline tags like [laughs] and [whispers], and generate natural speech across MSA, dialects, and Arabic-English code-switching — straight from raw text, no phonemizer or G2P. Ships with a set of natural synthetic voices. The Flash and Turbo tiers are free to download, fine-tune, and self-host. Built for voice agents, media, education, and accessibility.
The inline tags actually work without sounding robotic, which caught me off guard for Arabic TTS. Code-switching handled my mixed Arabic-English sentences surprisingly well too.
@burcubitgen “Caught me off guard“ might be our favorite kind of feedback. 🙏 Natural-sounding inline tags and real code-switching were exactly the two things we refused to compromise on for Arabic.
Thank you for testing it properly — keep the tough sentences coming!
@burcubitgen Good news! A new version is on the way, with better long-form text handling and more stable performance overall. Hope you'll come back and give it a try when it lands!
The code-switching between Arabic and English in the same sentence actually sounded natural, which caught me off guard. The inline tags like [whispers] feel like a really clean way to handle expressiveness without extra pipelines.
@muncusunlu67321 Really appreciate this. 🙏
Getting code-switching to sound natural in-sentence was hard-won, and the inline tags were a deliberate choice; we wanted expressiveness built into the model, not stitched on with extra pipelines.
Great to see that land with someone who clearly gets it. What else are you testing it on?
@muncusunlu67321 Good news! A new version is on the way, with better long-form text handling and more stable performance overall. Hope you'll come back and give it a try when it lands!
finally a tts that handles arabic dialects without sounding robotic, the [whispers] tag actually feels natural and the code-switching into english is pretty smooth too
@defneakkuzey This means a lot. 🙏 Dialects that don't sound robotic, natural expression tags, and smooth code-switching — those were exactly our priorities.
And there's more coming: we're working on a new version with stronger long-form text handling and improved overall stability.
Thank you for giving it a real test; would love to have you back to try the next version!