Open.Unmix.App – Towards Audio Separation On The Edge
summary
A web port of the open-unmix model runs entirely in the browser after conversion to TensorFlow.js, taking seconds to a few minutes per song to extract vocals and accompaniment and produce downloadable stems. The demo shows realtime mixing of isolated tracks, waveform zoom and color customization, and a workflow that publishes stems via cloud links into an embeddable multitrack player so shared URLs reproduce the exact mix and player configuration.
AI-written summary · about
related talks
51mThe Merger of Machine Learning, Music and Programming
14mThe nuts and bolts of music source separation
21mMusical Deep Neural Networks in the Browser
28mMusic and AI in the Browser with TensorFlow js and Magenta js
19mMixing voice & data
19mAudio Streaming - Using WebRTC for building your own Voice AIs
01h 14mExploring the hidden potential of sound data
55mAI Music Generation, From Prompt to Production