Kyrgyz Startup Unveils AI Speech Synthesis Model at CES 2026

Виктор Сизов Economy
VK X OK WhatsApp Telegram


The main product presented at the exhibition was KaniTTS — an open-source speech synthesis model. The technology's authors claim that it can generate speech in real-time three times faster and ten times cheaper than similar solutions from giants like ElevenLabs, OpenAI, and Google. The model is available for use under the Apache 2.0 license, making it free for developers.

KaniTTS boasts impressive technical specifications: it can create 15 seconds of audio recording in just one second, using a standard NVIDIA RTX 5080 graphics card. This allows for the integration of the technology without the need for expensive cloud solutions. The model has been downloaded over 15,000 times on the Hugging Face platform and supports eight languages, including Kyrgyz, English, German, and Chinese.

The second product presented was the Kyrgyz Whisper model, designed for automatic speech recognition. It was fine-tuned based on OpenAI's solution and uses data from 2000 hours of Kyrgyz speech, which reduced the recognition error rate from nearly 100% to 0.2%. This significantly improves support for languages that are underrepresented on the international stage.

NineNineSix's participation in the exhibition was organized by the High Technology Park of the Kyrgyz Republic. According to the PVT, the IT sector of Kyrgyzstan is showing significant success: over the past five years, service exports have increased 45 times. In 2024, local specialists earned $130 million in foreign markets, with 40% of this volume (over $50 million) coming from the USA.
VK X OK WhatsApp Telegram

Read also:

Write a comment: