Bishkek, Feb.19, 2026. /Kabar/. Kyrgyz developers have once again demonstrate professionalism on the international technology scene, the High-Tech Park (HTP) of Kyrgyzstan reports.
NineNineSix startup team presented a new version of their speech synthesis model, Kani TTS 2, which has already become one of the most popular TTS models on the Hugging Face platform, the world's largest catalog of artificial intelligence models.
Kani TTS 2 is a logical continuation of the team's previous developments and represents a significant technological leap: the model is now capable of reliably generating up to 40 seconds of live speech in a single run, more than double the capabilities of the first version.
The HTP noted that for a fully open-source model from Kyrgyzstan, reaching the top three TTS models on Hugging Face is a rare and significant achievement.
Who are NineNineSix?
NineNineSix is a Kyrgyzstani AI development team known for its projects in language technology.
The team previously released the first version of Kani TTS and developed a voice speaker and AI assistant, AkylAi, which became the first AI to speak Kyrgyz.
Voice for Underrepresented Languages
Most large AI companies focus on English and other global languages. Resource-poor languages are often left behind by technological advances. NineNineSix took a different approach.
Kani TTS 2 already supports English, Spanish, and Kyrgyz, and the model's architecture allows it to be trained for any other languages, accents, and dialects.
A unique feature of the project is that the team has published the full pre-training code. This means any country or research group can create their own voice model based on Kani TTS 2.
"Kani TTS 2 is the next step after our first version: we've made speech generation more stable and taught the model to produce longer fragments. We focus on compact and open models—they're easier to launch and adapt to different languages and accents, including resource-poor ones. It's important for us to demonstrate that world-class technology can be created in Kyrgyzstan, so we've open-sourced not only the model weights but also the entire pretraining code—so any team can train a TTS from scratch for their language," noted nineninesix.ai co-founder Nursultan Bakashov.
Kani TTS 2 introduces key improvements:
* Up to 40 seconds of stable speech generation per run;
* Support for zero-shot voice cloning—voice cloning based on a short audio fragment;
* Completely open architecture and training code;
* Ranked among the top 3 TTS models on Hugging Face.
HTP added that the model contains approximately 400 million parameters, was pre-trained on approximately 10,000 hours of speech data, and can run on GPUs with approximately 3 GB of video memory, making it suitable for both local and server use.
"Kani TTS 2 is not just another AI model. It is proof that Kyrgyzstani specialists are capable of creating world-class technologies and competing in the global artificial intelligence market. NineNineSix demonstrates that Kyrgyzstan can be not only a consumer but also a creator of advanced AI solutions," the HTP reported.
