A portable robot audition software system for multiple simultaneous speech signals

Hiroshi G. Okuno, S. Yamamoto, K. Nakadai, J. M. Valin, Tetsuya Ogata, K. Komatani

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Since a robot is deployed in various kinds of environments, the robot audition system should work with minimum prior information on environments to localize, separate and recognize utterances by multiple simultaneous talkers. For example, it should not assume either the number of speakers, the location of speakers for sound source separation (SSS), or specially tuned acoustic model for automatic speech recognition (ASR). We developed \HARK" portable robot audition that uses eight microphones installed on the surface of robot's body such as Honda ASIMO, and SIG-2 and Robovie-R2 at Kyoto University. HARK integrates SSS and ASR by using the Missing-Feature Theory. For SSS, we use Geometric Source Separation and multi-channel post-filter to separate each utterance. Since separated speech signals are distorted due to interfering talkers and sound source separation, multi-channel post-filter enhanced speech signals. At this process, we create a missing feature mask that specifies which acoustic features are reliable in time-frequency domain. Multi-band Julius, a missing-feature-theory based ASR, uses this mask to avoid the inuence of unreliable features in recognizing such distorted speech signals. The system demonstrated a waitress robot that accepts meal orders placed by three actual human talkers.

Original languageEnglish
Title of host publicationProceedings - European Conference on Noise Control
Pages483-488
Number of pages6
Publication statusPublished - 2008
Externally publishedYes
Event7th European Conference on Noise Control 2008, EURONOISE 2008 - Paris, France
Duration: 2008 Jun 292008 Jul 4

Other

Other7th European Conference on Noise Control 2008, EURONOISE 2008
CountryFrance
CityParis
Period08/6/2908/7/4

Fingerprint

Source separation
Audition
robots
Hearing
Software
Robots
computer programs
Speech recognition
acoustics
Acoustic waves
speech recognition
Masks
Acoustics
masks
filters
Microphones
microphones
Meals

ASJC Scopus subject areas

  • Acoustics and Ultrasonics
  • Public Health, Environmental and Occupational Health
  • Building and Construction
  • Mechanical Engineering
  • Industrial and Manufacturing Engineering
  • Automotive Engineering
  • Aerospace Engineering

Cite this

Okuno, H. G., Yamamoto, S., Nakadai, K., Valin, J. M., Ogata, T., & Komatani, K. (2008). A portable robot audition software system for multiple simultaneous speech signals. In Proceedings - European Conference on Noise Control (pp. 483-488)

A portable robot audition software system for multiple simultaneous speech signals. / Okuno, Hiroshi G.; Yamamoto, S.; Nakadai, K.; Valin, J. M.; Ogata, Tetsuya; Komatani, K.

Proceedings - European Conference on Noise Control. 2008. p. 483-488.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Okuno, HG, Yamamoto, S, Nakadai, K, Valin, JM, Ogata, T & Komatani, K 2008, A portable robot audition software system for multiple simultaneous speech signals. in Proceedings - European Conference on Noise Control. pp. 483-488, 7th European Conference on Noise Control 2008, EURONOISE 2008, Paris, France, 08/6/29.
Okuno HG, Yamamoto S, Nakadai K, Valin JM, Ogata T, Komatani K. A portable robot audition software system for multiple simultaneous speech signals. In Proceedings - European Conference on Noise Control. 2008. p. 483-488
Okuno, Hiroshi G. ; Yamamoto, S. ; Nakadai, K. ; Valin, J. M. ; Ogata, Tetsuya ; Komatani, K. / A portable robot audition software system for multiple simultaneous speech signals. Proceedings - European Conference on Noise Control. 2008. pp. 483-488
@inproceedings{265b13213ef84e51a0ba05752150a1b9,
title = "A portable robot audition software system for multiple simultaneous speech signals",
abstract = "Since a robot is deployed in various kinds of environments, the robot audition system should work with minimum prior information on environments to localize, separate and recognize utterances by multiple simultaneous talkers. For example, it should not assume either the number of speakers, the location of speakers for sound source separation (SSS), or specially tuned acoustic model for automatic speech recognition (ASR). We developed \HARK{"} portable robot audition that uses eight microphones installed on the surface of robot's body such as Honda ASIMO, and SIG-2 and Robovie-R2 at Kyoto University. HARK integrates SSS and ASR by using the Missing-Feature Theory. For SSS, we use Geometric Source Separation and multi-channel post-filter to separate each utterance. Since separated speech signals are distorted due to interfering talkers and sound source separation, multi-channel post-filter enhanced speech signals. At this process, we create a missing feature mask that specifies which acoustic features are reliable in time-frequency domain. Multi-band Julius, a missing-feature-theory based ASR, uses this mask to avoid the inuence of unreliable features in recognizing such distorted speech signals. The system demonstrated a waitress robot that accepts meal orders placed by three actual human talkers.",
author = "Okuno, {Hiroshi G.} and S. Yamamoto and K. Nakadai and Valin, {J. M.} and Tetsuya Ogata and K. Komatani",
year = "2008",
language = "English",
pages = "483--488",
booktitle = "Proceedings - European Conference on Noise Control",

}

TY - GEN

T1 - A portable robot audition software system for multiple simultaneous speech signals

AU - Okuno, Hiroshi G.

AU - Yamamoto, S.

AU - Nakadai, K.

AU - Valin, J. M.

AU - Ogata, Tetsuya

AU - Komatani, K.

PY - 2008

Y1 - 2008

N2 - Since a robot is deployed in various kinds of environments, the robot audition system should work with minimum prior information on environments to localize, separate and recognize utterances by multiple simultaneous talkers. For example, it should not assume either the number of speakers, the location of speakers for sound source separation (SSS), or specially tuned acoustic model for automatic speech recognition (ASR). We developed \HARK" portable robot audition that uses eight microphones installed on the surface of robot's body such as Honda ASIMO, and SIG-2 and Robovie-R2 at Kyoto University. HARK integrates SSS and ASR by using the Missing-Feature Theory. For SSS, we use Geometric Source Separation and multi-channel post-filter to separate each utterance. Since separated speech signals are distorted due to interfering talkers and sound source separation, multi-channel post-filter enhanced speech signals. At this process, we create a missing feature mask that specifies which acoustic features are reliable in time-frequency domain. Multi-band Julius, a missing-feature-theory based ASR, uses this mask to avoid the inuence of unreliable features in recognizing such distorted speech signals. The system demonstrated a waitress robot that accepts meal orders placed by three actual human talkers.

AB - Since a robot is deployed in various kinds of environments, the robot audition system should work with minimum prior information on environments to localize, separate and recognize utterances by multiple simultaneous talkers. For example, it should not assume either the number of speakers, the location of speakers for sound source separation (SSS), or specially tuned acoustic model for automatic speech recognition (ASR). We developed \HARK" portable robot audition that uses eight microphones installed on the surface of robot's body such as Honda ASIMO, and SIG-2 and Robovie-R2 at Kyoto University. HARK integrates SSS and ASR by using the Missing-Feature Theory. For SSS, we use Geometric Source Separation and multi-channel post-filter to separate each utterance. Since separated speech signals are distorted due to interfering talkers and sound source separation, multi-channel post-filter enhanced speech signals. At this process, we create a missing feature mask that specifies which acoustic features are reliable in time-frequency domain. Multi-band Julius, a missing-feature-theory based ASR, uses this mask to avoid the inuence of unreliable features in recognizing such distorted speech signals. The system demonstrated a waitress robot that accepts meal orders placed by three actual human talkers.

UR - http://www.scopus.com/inward/record.url?scp=84874864805&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84874864805&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84874864805

SP - 483

EP - 488

BT - Proceedings - European Conference on Noise Control

ER -