The Audacity audio software program now includes new OpenVINO AI effects for users on Windows. The new features include Noise Suppression and Transcription useful for podcasts, and Music Generation, Music Style Remix, and Music Separation, useful for music.
OpenVINO (Open Visual Inference and Neural network Optimization) is a toolkit that Intel released that allows deep learning models to run locally on Intel hardware. Each of these new features that leverage OpenVINO use AI that runs directly on your hardware.
Explaining what each of the features do, Audacity writes:
- The Noise Suppression does what it says on the tin - it suppresses noise. As such it behaves similar to Audacity’s built-in Noise Removal effect.
- The Transcription powered by Whisper.cpp can both transcribe and translate words and outputs to a label track. If you want to export these transcriptions, you can do so via File → Export Other → Export Labels.
- The Transcription powered by Whisper.cpp can both transcribe and translate words and outputs to a label track. If you want to export these transcriptions, you can do so via File → Export Other → Export Labels.
- Music Generation and Music Style Remix use Stable Diffusion (and Riffusion in particular) to generate new music from a prompt, or based on pre-existing music, respectively.
- Music Separation can split a song into either it’s vocal and instrumental parts, or into vocals, drums, bass and a combined “anything else” part. This is ideal for creating covers and playalongs.
If you want to learn more about the features in detail, you can use the following links:
To use these OpenVINO plugins, which were developed by Intel, you need to download them from Intel’s dedicated GitHub page. Audacity does note that you may be able to compile the software from source on Linux and macOS but that no instructions have been provided.
Audacity is a popular open source audio editing software that is a popular choice, especially in Linux circles where it’s one of the main options for editing audio. As mentioned earlier, these AI tools will need to be compiled manually on Linux installations. If it can be done, it’s likely that someone in the Linux community will compile them and make them available through various channels.
Source: Audacity