Every day, countless devices process huge amounts of digital audio data - video calls, livestreams, music streams, etc. All of this audio undergoes countless audio processing transformations to provide the best audio experience. For real time applications, this is accomplished through block audio processing.
Each sampling period, complex math is performed on blocks of audio samples.
Significant digital signal processing expertise is required to design reliable block processing audio systems.
Development of stable products requires weeks or months of traditional software development.
Current digital signal processing systems are restricted to linear solutions you can code.
High Quality DNN Audio Processing Models
Real-Time Audio Machine Learning Backend
Our models learn complex, real-world non-linear audio transformations.
Our optimized C++ ML backend allows us to use our models at professional sample rates in real time.
All we need to train new models is an input dataset, a target dataset, and a few hours of training time.
We are developing commercial audio processings plugins and optimized processing libraries.
Duncan is our co-founder and product lead. As an audio researcher, musician, and recording engineer, his goal is to empower musicians.
Ladan is our co-founder and tech lead. She is a super motivated ML engineer who lives and breathes network design, audio datasets, and optimization.
Jesus is our real-time audio expert. He is a musician and electrical engineer who has built some amazing digital music products.. and is building a few more as we speak!