Exhibition Program

Science of Media Information

15

Voice command and speech communication in car

- World’s best voice capture and recognition technologies -

Abstract

Our technology can support a speech command and hand-free communication even in noisy environment such as road noise without any stresses. Clear speech can be picked up from the noise-mixed sound in order to realize speech command with high accuracy. A lot of computational complexy and memory was required to keep speech quality and reduce only noise so far. This problem can be solved by using our acoustical knowhow, moreover, low latency was able to be also achieved. In addition, a sign of howling was able to be detected rapidly by combining multiple microphone array. Our goal is to improve an in-car acoustical environment by reducing noises which are road noise, engiine noise, and any sound from other cars. We will also try to establish an event detection technology in order to help a driving assistant or an early maintenance by detecting emergency car or anomalous in sound.

References

  • [1] Y. Hioka, K. Furuya, K. Kobayashi, K. Niwa, Y. Haneda, “Underdetermined sound source separation using power spectrum density estimated by combination of directivity gain,” IEEE Trans. Audio, Speech, Lang. Process., vol. 21, no. 6, pp. 1240-1250, 2013.
  • [2] T. Yoshioka, N. Ito, M. Delcroix, A. Ogawa, K. Kinoshita, M. Fujimoto, C. Yu, W. H. Fabian, M. Espi, T. Higuchi, S. Araki, T. Nakatani, “The NTT CHiME-3 system: Advances in speech enhancement and recognition for mobile multi-microphone devices,” in Proc. IEEE ASRU, pp. 436-443, Dec. 2015.

Poster

Photos

Contact

Noboru Harada, Media intelligence laboratory
Email: