To provide a parameter estimating technique which allows more robust parameter estimation to be achieved by obtaining an error in a smaller unit on the basis of a fine-grained error standard and minutely considering "a degree of incorrectness" within the incorrectness.
A parameter estimating device includes: a recording section for storing a probability finite state model containing a parameter for use in speech recognition, learning data, a correct state transition sequence which is a state transition sequence corresponding to a correct speech recognition result of the learning data, and an adjustment parameter which is a parameter for adjusting the probability finite state model; a recognizing section for generating a recognition state transition sequence which is a state transition sequence corresponding to a speech recognition result obtained as a result of performing speech recognition with respect to the learning data using the probability finite state model; a fine-grained error measure calculating section for calculating an error measure on the basis of a difference between the correct state transition sequence and the recognition state transition sequence; and a parameter estimating section for correcting the adjustment parameter according to the error measure.
HORI TAKAAKI
NAKAMURA ATSUSHI
JP2004279701A | 2004-10-07 | |||
JP2010250161A | 2010-11-04 | |||
JP2011164336A | 2011-08-25 |
WO2010019831A1 | 2010-02-18 |
JPN6014044390; S.KAPADIA, et al.: '"MMI TRAINING FOR CONTINUOUS PHONEME RECOGNITION ON THE TIMIT DATABASE"' Proceedings of the 1993 IEEE International Conference on Acoustics, Speech and Signal Processing (IC Vol.II, 199304, pp.491-494
JPN6011028277; 大庭隆伸他: '"識別的誤り訂正学習における対立単語列と素性の選定"' 電子情報通信学会技術研究報告 Vol.107,No.406, 200712, pp.235-240
JPN6014044390; S.KAPADIA, et al.: '"MMI TRAINING FOR CONTINUOUS PHONEME RECOGNITION ON THE TIMIT DATABASE"' Proceedings of the 1993 IEEE International Conference on Acoustics, Speech and Signal Processing (IC Vol.II, 199304, pp.491-494
JPN6014044383; Daniel POVEY, et al.: '"MINIMUM PHONE ERROR AND I-SMOOTHING FOR IMPROVED DISCRIMINATIVE TRAINING"' Proceedings of the 2002 IEEE International Conference on Acoustics, Speech, and Signal Processing (I Vol.I, 200205, pp.105-108
JPN6014044386; Daniel POVEY, et al.: '"BOOSTED MMI FOR MODEL AND FEATURE-SPACE DISCRIMINATIVE TRAINING"' Proceedings of the 2008 IEEE International Conference on Acoustics, Speech and Signal Processing (IC , 200803, pp.4057-4060
JPN6014044388; Erik MCDERMOTT, et al.: '"DISCRIMINATIVE TRAINING BASED ON AN INTEGRATED VIEW OF MPE AND MMI IN MARGIN AND ERROR SPACE"' Proceedings of the 2010 IEEE International Conference on Acoustics Speech and Signal Processing (ICA , 201003, pp.4894-4897
CSNG200800120035; 大庭隆伸他: '"識別的誤り訂正学習における対立単語列と素性の選定"' 電子情報通信学会技術研究報告 Vol.107,No.406, 200712, pp.235-240
JPN6014044383; Daniel POVEY, et al.: '"MINIMUM PHONE ERROR AND I-SMOOTHING FOR IMPROVED DISCRIMINATIVE TRAINING"' Proceedings of the 2002 IEEE International Conference on Acoustics, Speech, and Signal Processing (I Vol.I, 200205, pp.105-108
JPN6014044386; Daniel POVEY, et al.: '"BOOSTED MMI FOR MODEL AND FEATURE-SPACE DISCRIMINATIVE TRAINING"' Proceedings of the 2008 IEEE International Conference on Acoustics, Speech and Signal Processing (IC , 200803, pp.4057-4060
JPN6014044388; Erik MCDERMOTT, et al.: '"DISCRIMINATIVE TRAINING BASED ON AN INTEGRATED VIEW OF MPE AND MMI IN MARGIN AND ERROR SPACE"' Proceedings of the 2010 IEEE International Conference on Acoustics Speech and Signal Processing (ICA , 201003, pp.4894-4897
CSNG200800120035; 大庭隆伸他: '"識別的誤り訂正学習における対立単語列と素性の選定"' 電子情報通信学会技術研究報告 Vol.107,No.406, 200712, pp.235-240
Yukio Nakamura
Yoshimura Munehiro
Next Patent: DRIVE CONTROL DEVICE, DISPLAY DEVICE EQUIPPED WITH THE SAME, AND DRIVE CONTROL METHOD