Improvement of robot audition by interfacing sound source separation and automatic speech recognition with Missing Feature Theory

2004 ◽  
Author(s):  
S. Yamamoto ◽  
K. Nakadai ◽  
H. Tsujino ◽  
T. Yokoyama ◽  
H.G. Okuno
Keyword(s):  
Speech Recognition ◽  
Automatic Speech Recognition ◽  
Sound Source ◽  
Source Separation ◽  
Sound Source Separation ◽  
Robot Audition ◽  
Missing Feature Theory ◽  
Missing Feature ◽  
Feature Theory

scholarly journals Simultaneous Speech Recognition Based on Automatic Missing Feature Mask Generation by Integrating Sound Source Separation

2007 ◽  
Vol 25 (1) ◽  
pp. 92-102
Author(s):  
Shunichi Yamamoto ◽  
Jean-Marc Valin ◽  
Kazuhiro Nakadai ◽  
Mikio Nakano ◽  
Hiroshi Tsujino ◽  
...  
Keyword(s):  
Speech Recognition ◽  
Sound Source ◽  
Source Separation ◽  
Sound Source Separation ◽  
Missing Feature

scholarly journals Noise Robust Automatic Speech Recognition Method for the Robot with Motor Noise using Missing Feature Theory

2007 ◽  
Vol 25 (8) ◽  
pp. 1189-1198
Author(s):  
Yoshitaka Nishimura ◽  
Mitsuru Ishizuka ◽  
Kazuhiro Nakadai ◽  
Mikio Nakano ◽  
Hiroshi Tsujino
Keyword(s):  
Speech Recognition ◽  
Automatic Speech Recognition ◽  
Recognition Method ◽  
Motor Noise ◽  
Noise Robust ◽  
Missing Feature Theory ◽  
Missing Feature ◽  
Feature Theory

scholarly journals Soft missing-feature mask generation for Robot Audition

2010 ◽  
Vol 1 (1) ◽  
Author(s):  
Toru Takahashi ◽  
Kazuhiro Nakadai ◽  
Kazunori Komatani ◽  
Tetsuya Ogata ◽  
Hiroshi G. Okuno
Keyword(s):  
Human Robot Interaction ◽  
Sigmoid Function ◽  
Robot Interaction ◽  
Time Frequency ◽  
Sound Source Separation ◽  
Robot Audition ◽  
Frequency Components ◽  
Interaction Task ◽  
Missing Feature Theory ◽  
Missing Feature

AbstractThis paper describes an improvement in automatic speech recognition (ASR) for robot audition by introducing Missing Feature Theory (MFT) based on soft missing feature masks (MFM) to realize natural human-robot interaction. In an everyday environment, a robot’s microphones capture various sounds besides the user’s utterances. Although sound-source separation is an effective way to enhance the user’s utterances, it inevitably produces errors due to reflection and reverberation. MFT is able to cope with these errors. First, MFMs are generated based on the reliability of time-frequency components. Then ASR weighs the time-frequency components according to the MFMs. We propose a new method to automatically generate soft MFMs, consisting of continuous values from 0 to 1 based on a sigmoid function. The proposed MFM generation was implemented for HRP-2 using HARK, our open-sourced robot audition software. Preliminary results show that the soft MFM outperformed a hard (binary) MFM in recognizing three simultaneous utterances. In a human-robot interaction task, the interval limitations between two adjacent loudspeakers were reduced from 60 degrees to 30 degrees by using soft MFMs.

Sign in / Sign up

Export Citation Format

Share Document