CLMR on ONNX Runtime

On this page, you are running a CLMR model on the ONNX Runtime in your browser.
No data is sent to a server, all predictions are done on your device!

In short, the following happens under the hood:

It will first load a pre-trained model in the background.
The model extracts audio representations from raw samples from the audio buffer.
The audio representations are given to a linear classifier, which predicts the corresponding audio tags.
The multi-label predictions are shown in the bar plot.

Select a piece of music:

Or select a local file:

No data is uploaded to the server.
Ideally, the sample rate of the file should be 22,050Hz. But smaller / higher sample rates also work.

Loading ONNX model...