A system that hears and thinks like humans.

What we do

We are surrounded by millions of different sounds that contain important clues about our surroundings. For example, if you hear someone screaming, you know that there is an emergency. If someone coughs a lot, it means that he or she is sick. How great would it be if AI could understand what all these “sounds” imply? delivers a top-quality machine listening technology to solve issues and challenges around the world. We provide our technology through accessible cloud API and Edge SDKs, which would add hearing abilities to any devices or applications.

Emergency detection

Whereas cameras are widely used for monitoring purposes, their limitation in leaving blind spots remains unresolved. Clues of emergency are often clearer in sounds, such as screaming or glass breaking, and these events can be detected by our machine listening AI. The automated monitoring system is more efficient as it is cheaper than human labor while eradicating human errors.


Humans intrinsically understand the context of their environment through sound, but it is still difficult to do so for computers. Today, we control many devices with touch screens or voice commands, but the next generation of AI system should be able to pick up what’s happening simply through sound to automatically take suitable action in appropriate times.

Music analysis

Music is a unique form of audio, containing lots of information regarding genre, mood, key, and tempo; however, metadata is not rich enough to describe music. Extracted music information can be potentially used for contents-based music recommendation or searching and grouping huge amounts of music clips.

Speech analysis

Although at the moment only speech recognition technology is widely available, we think that speech contains much more than simply conveying meaning. Through the tone of the voice, humans are able to understand non-verbal information such as age, gender, illness, and emotion, and we are striving for AI to do the same.

Anomaly detection

A machine usually produces an awkward sound when it is not working properly or about to break down. But it is often hard to analyze the problem because we cannot fix the machine while it is operating. Anomaly detection technology can solve this problem with better accuracy and consistency.

Audio search

Today, searching for information on the internet through text queries and images are widely available. What if we want to search for an audio clip? Using our technology, it is possible to search for a certain scene or event in a video or audio clip, or sort and organize them just like you would for texts or images.

Cutting-edge sound AI research

The team achieved top ranks in all tasks of the IEEE Detection and Classification of Acoustic Scenes and Events (DCASE) Challenge 2017, the most prestigious competition available in the machine listening field. In 2018, on the DCASE General purpose audio tagging, our team ranked first among 558 teams on Kaggle.