GIST · Audio Intelligent Technology & Research Lab

AI that understands and creates sound

We study how machines recognize, analyze, and generate speech and audio.

Research

Eight research areas

Our research spans recognizing and analyzing sound through to generating it.

Transcribes real-world speech despite noise and accents.

Tells who spoke when, even through overlapping speech.

Identifies the scene and what's happening from surrounding sound.

Generates natural speech and audio from text.

Removes noise and reverberation to restore clear speech.

Compresses audio into fewer bits while preserving quality.

Recreates the direction and space of sound.

Forecasts climate change from time-series data.

Updates