Skip to content

HeAR

A bioacoustic model that reads cough sounds

Visit official site

What HeAR is

The model is a Google Research bioacoustic foundation model that looks for early signs of disease in body sounds.

What you can do with it

  • Request access to the API
  • Analyze cough or other audio
  • Review possible disease indications

Key features

  • Health related sound patterns
  • Custom bioacoustic model building
  • Lower data and compute needs

Pros & cons

✓ What we like

  • Generalization across microphones is noted
  • Reduced data, setup and compute needs

! What to watch out for

  • Researchers must request API access
  • Analysis needs cough or bioacoustic audio

FAQ

How much training audio?

The material states 300 million samples, about 100 million coughs.

Who developed it?

The material names the Google Research team.

Last reviewed: 2026-11-15

More AI audio tools tools

View all →

How we review