First, I want to say a sincere thank you to ServeurpersoCom, the developer of omnivoice.cpp. This project made it possible for me to use OmniVoice through a practical local C++ runtime, and it has become an important runtime layer for the OmniVoice workflow inside my project.
Today, I’m introducing LA Studio, short for Local Audio Studio.
https://github.com/dduongtrandai/LA-Studio
LA Studio is an offline AI audio desktop application built for people who want to experiment with modern speech AI without depending on cloud APIs. The goal is simple: bring speech-to-text, text-to-speech, voice cloning, voice design, model downloads, runtime management, and audio preview tools into one native desktop app.
Instead of sending your audio, prompts, or generated voices to external services, LA Studio is designed around local inference. It is built for creators, developers, researchers, and anyone who wants more privacy and control when working with AI audio models.
With LA Studio, I want to make local AI audio easier to try, compare, and use. The app brings together multiple open speech model families, including Whisper, Qwen3-ASR, Kokoro, Qwen3-TTS, VibeVoice, VieNeu-TTS, VoxCPM2, and OmniVoice. For OmniVoice workflows, omnivoice.cpp provides a powerful foundation for local text-to-speech, voice cloning, and voice design.
The project is still evolving, but the vision is clear: a practical offline AI audio workstation where users can browse models, download the required files, validate runtimes, check hardware compatibility, generate or transcribe audio, preview results, and manage everything from a single interface.
I’m especially interested in feedback from people who work with:
- local AI speech tools
- text-to-speech and voice cloning
- multilingual audio generation
- podcasting, narration, accessibility, and content creation
- C++/Qt desktop applications
- model packaging and runtime integration
If you care about private, local-first AI audio workflows, I would love for you to try LA Studio, share feedback, report issues, and help shape the direction of the project.
LA Studio is my attempt to make powerful open speech models more accessible from a real desktop app, not just scripts and command-line experiments. My hope is that it becomes a useful bridge between open AI audio research and everyday creative workflows.
Thanks again to the open-source developers behind these models and runtimes, especially omnivoice.cpp, for making projects like this possible.
First, I want to say a sincere thank you to ServeurpersoCom, the developer of
omnivoice.cpp. This project made it possible for me to use OmniVoice through a practical local C++ runtime, and it has become an important runtime layer for the OmniVoice workflow inside my project.Today, I’m introducing LA Studio, short for Local Audio Studio.
https://github.com/dduongtrandai/LA-Studio
LA Studio is an offline AI audio desktop application built for people who want to experiment with modern speech AI without depending on cloud APIs. The goal is simple: bring speech-to-text, text-to-speech, voice cloning, voice design, model downloads, runtime management, and audio preview tools into one native desktop app.
Instead of sending your audio, prompts, or generated voices to external services, LA Studio is designed around local inference. It is built for creators, developers, researchers, and anyone who wants more privacy and control when working with AI audio models.
With LA Studio, I want to make local AI audio easier to try, compare, and use. The app brings together multiple open speech model families, including Whisper, Qwen3-ASR, Kokoro, Qwen3-TTS, VibeVoice, VieNeu-TTS, VoxCPM2, and OmniVoice. For OmniVoice workflows,
omnivoice.cppprovides a powerful foundation for local text-to-speech, voice cloning, and voice design.The project is still evolving, but the vision is clear: a practical offline AI audio workstation where users can browse models, download the required files, validate runtimes, check hardware compatibility, generate or transcribe audio, preview results, and manage everything from a single interface.
I’m especially interested in feedback from people who work with:
If you care about private, local-first AI audio workflows, I would love for you to try LA Studio, share feedback, report issues, and help shape the direction of the project.
LA Studio is my attempt to make powerful open speech models more accessible from a real desktop app, not just scripts and command-line experiments. My hope is that it becomes a useful bridge between open AI audio research and everyday creative workflows.
Thanks again to the open-source developers behind these models and runtimes, especially
omnivoice.cpp, for making projects like this possible.