OpenVINO AI Plugins for Audacity
OpenVINO AI Plugins for Audacity is a remarkable collection of AI-driven effects, generators, and analyzers designed to integrate seamlessly with Audacity, the popular audio editing software. This project enables users to perform an array of sophisticated audio processing tasks directly on their local devices without the need for an internet connection. Leveraging the power of OpenVINO, it runs AI models effectively on the user's CPU, GPU, or NPU.
Features
The project offers several advanced audio manipulation features:
-
Music Separation: This feature allows users to deconstruct a mono or stereo audio track into distinct stems, such as drums, bass, vocals, and other instruments. It gives users the ability to isolate specific musical elements for more detailed editing.
-
Noise Suppression: It provides an effective way of removing unwanted background noise from audio samples, ensuring cleaner and more professional sounding recordings.
-
Music Generation & Continuation: Through the use of the MusicGen LLM, users can create musical snippets or extend already existing segments of music. This feature is perfect for creative explorers looking to generate new compositions or seamlessly transition between musical ideas.
-
Whisper Transcription: Using whisper.cpp, this tool can transcribe or translate spoken audio or vocals into text, generating a label track in Audacity. It's an invaluable aid for tasks requiring annotated audio files.
Installation
To start using OpenVINO AI Plugins for Audacity, users can find installation packages and detailed instructions for Windows on the project's release page.
Build Instructions
For those interested in customizing or building the plugins, comprehensive build instructions are available for both Windows and Linux platforms.
Participation and Community
The community around OpenVINO AI Plugins for Audacity is active and inviting. Users are encouraged to submit issues or feedback through the project's GitHub issue tracker. This can include questions, bug reports, suggestions for new features, and general feedback on the project. Additionally, contributions are always welcomed, whether large or small, with an open invitation to submit pull requests.
Acknowledgements
The project's success is supported by contributions from various sources:
- The Audacity development team and Muse Group have been instrumental in providing support and collaboration.
- The Whisper transcription tool benefits from the brilliant work on whisper.cpp, with integration via the OpenVINO backend.
- Music Generation utilizes Meta's MusicGen model, specifically the MusicGen-Small and MusicGen-Small-Stereo versions, originally from Hugging Face.
- Music Separation employs the Demucs v4 model, another product from Meta, adeptly adapted to function with OpenVINO.
- Noise Suppression draws on multiple models, including those from OpenVINO’s Open Model Zoo and the DeepFilterNet series, which have been adapted for this project.
Inspirations
The team continually learns and adopts best practices from OpenVINO Notebooks, which provide a wealth of information and inspiration for implementing effective AI pipelines.
OpenVINO AI Plugins for Audacity represents an exciting leap forward in making advanced audio processing accessible to all users, harnessing the power of AI to elevate creative possibilities within the Audacity software.