About
NeuralSound is an AI music source separation app. You upload a song or a video, and it splits the audio into separate parts so you can keep the piece you need and leave the rest.
Most people come with one of three jobs. If you want to sing or play over a song, the vocal remover takes the lead singer out and keeps the full backing track for karaoke, covers and practice. If you want the voice on its own, the instrumental remover takes the band away instead, so you can learn a melody or start a remix. And if you recorded a vlog, lesson, interview or demo with music underneath, the background music remover is tuned for speech, so it pulls the music out while keeping your words clear.
For more control, music separation splits a track into vocals, drums, bass and other parts. NeuralSound also creates time-synced lyrics and captions in more than 96 languages and can turn them into karaoke videos. We published a small benchmark on five MUSDB18-HQ tracks, with the method and numbers on our site, so you can check the quality for yourself.
It works in the browser with MP3, WAV, FLAC, M4A, MP4 and MOV files, and it's on mobile too: NeuralSound on Google Play for Android and NeuralSound on the App Store for iPhone and iPad. Feedback is welcome at support@neuralsound.org.