Toshiba AI Technology Catalog

  • Anomaly detection

Audio abnormaly detection

Realize early detection of equipment abnormalies based on signs of degradation in operating sound.

  • Using an original Variational AutoEncoder method, it became possible to detect even faint sounds of deterioration without being affected by unknown noise.
  • Aiming for applications in the diagnosis of degradation in infrastructure facilities


  • Degradation diagnosis for equipment and facilities
  • Audio testing at manufacturing sites

Benchmarks, strengths, and track record

  • Accurately infer the degree of equipment degradation (Correlation coefficient: 0.99) even using audio signals overlaid with unknown noise. (Acoustical Society of Japan 2019 Autumn Meeting)


Please include the title “Toshiba AI Technology Catalog: Audio abnormality detection” or the URL in the inquiry text.
Please note that because this technology is currently the subject of R&D activities, immediate responses to inquiries may not be possible.


  • Hiroyuki Yanagihashi; “Study of early detection of device degradation signs using AutoEncoder”; Acoustical Society of Japan 2019 Autumn Meeting, 2019.