Source Separation
Discover a new world of sounds.
Discover a new world of sounds.
Explore limitless possibilities of sound
with Gaudio Lab’s source separation technology.
Explore limitless possibilities
of sound with Gaudio Lab’s
source separation technology.
Source separation refers to the extraction of individual sound sources from a mix of audio signals.
Examples include extracting an instrumental sound from a song, removing MR, and canceling noise.
GSEP (Gaudio source SEParation) is an AI source separation model developed by Gaudio Lab,
which has already proven to be the world's best in various external tests.
GSEP is sepcial for a reason
- 10.02*Amazing separation quality
We outperform Sony, Meta, and Deezer on signal-to-distortion ratio (SDR)*, a popular metric used in academia for measuring separation quality.
Signal-to-Distortion Ratio [dB]10987657.248.136.868.0810.02 - Fast separation with low
computational complexityThanks to its low computational complexity, GSEP can be used for
real-time noise canceling. Get clear voice without waiting -
Just Voice works real-time.Thanks to its low computational complexity, GSEP can be used for
real-time noise canceling. Get clear voice without waiting -
Just Voice works real-time. - Simple application
via Cloud API and SDKEasily integrate GSEP wherever you need it. We provide various features
in the form you desire, including clean stem separation, real-time noise removal,
and lyric sync for karaoke.
Gaudio Lab's technologies
are already widely used.
Korean music streaming platform NHN Bugs
We provided real-time lyric services.Korean music streaming platform VIBE
Met the user needs by improving the synchronization accuracy of lyrics to the melody.