Neural Sound

Inspiration

What it does

How I built it

Challenges I ran into

Accomplishments that I'm proud of

What I learned

What's next for Neural Sound

Inspiration

Music and audio creators were stuck choosing between expensive professional software and low quality free tools. We wanted to make studio grade source separation available to everyone.

What it does

Neural Sound uses AI music source separation to split any track into its parts. You get a vocal remover, instrumental remover, background music remover, vocal isolator and full music separation: https://neuralsound.org/vocal-remover https://neuralsound.org/instrumental-remover https://neuralsound.org/background-music-remover https://neuralsound.org/vocal-isolator https://neuralsound.org/tools/music-separation

How we built it

Deep neural networks trained on thousands of professionally mixed songs identify and isolate individual audio components — vocals, drums, bass and other instruments — from a full mix.

Challenges we ran into

Training accurate separation models and keeping inference fast enough for everyday users. We iterated on model quality, simplified the interface and shipped focused tools one at a time so each result stayed clean.

Accomplishments that we are proud of

Purpose built tools for vocal removal, instrumental removal, background music removal, vocal isolation and full stem separation, all from one platform.

What we learned

Start with a real problem you have felt yourself, ship a simple version early, and keep improving the core result quality.

What is next for Neural Sound

Expanding the toolset with more separation modes, faster processing and an API so developers can embed source separation in their own apps.

Built With

• ai
• audio
• machine-learning
• music
• python