StyleTTS2 UA Denys is an ONNX-exported version of a Ukrainian multispeaker text-to-speech model. The model is intended for on-device inference, such as use with ONNX Runtime on platforms like iOS. It is a direct export of the PyTorch weights from the original patriotyk/styletts2_ukrainian_multispeaker model, with only a minor graph modification to ensure compatibility with ONNX Runtime. No new training or data modifications were made in this export.
The repository credits the original model, training data, and Ukrainian language front-end to the developer @patriotyk, and encourages users to cite and support the original work. The front-end components referenced include modules for Ukrainian word stress and IPA transcription, specifically patriotyk/ukrainian-word-stress and patriotyk/ipa-uk. export with opset 19 and followed the procedures established in the original codebase.
The model is distributed under the MIT license, allowing for open use and modification. StyleTTS2 UA Denys is positioned as a resource for those seeking Ukrainian language text-to-speech capabilities, particularly for multispeaker scenarios and on-device applications.
StyleTTS2 UA Denys sits in PulseGate's Other AI category. It provides developers with an open-source Ukrainian text-to-speech model for on-device and local inference. StyleTTS2 UA Denys is an open-source project aimed at speech AI developers. The project is open source (MIT). It runs on the web and API.
Behind StyleTTS2 UA Denys is vasyadeva, and the product first shipped in 2025. The project is developed in the open on GitHub with 16 stars. Among its 5 catalogued features are text-to-speech, ukrainian language, and multispeaker support.
Latest indexed changes and source events
vasyadeva/StyleTTS2-UA-Denys verified by the PulseGate indexer
Other apps tracked under the same category.
