Toshiba AI Technology Catalog

  • Media transformation

Specified direction speech enhancement

Enhance and recognize only speech coming from the direction specified by the user.


  • Complex neural networks distinguish speech coming from a specified direction vs. environmental noise or speech coming from another direction.

Applications



  • Applications in smart speakers
  • Voice-operated home devices
  • Supports the creation of minutes in meetings with large numbers of participants

Benchmarks, strengths, and track record



  • Dramatically improves the accuracy of speech recognition when there is surrounding noise, compared to existing speech enhancement using deep neural networks

Inquiries



Please include the title “Toshiba AI Technology Catalog: Specified direction speech enhancement” 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.

References:

  • Daichi Hayakawa et.al.; “Fundamental study of specified direction speech recognition technology using complex neural network-based mask estimation”; Acoustical Society of Japan 2019 Autumn Meeting, pp.177-180, 2019.