Results 1 - 10 of 1,276

Table 3. HMM G2HMM

in A Probabilistic Approach for the Semantic Analysis
by J. Haas, J. Hornegger, E. Nöth, H. Niemann 1998
"... In PAGE 7: ... Table3 . Accuracy and number of errors of the experiments starting the count for the relative frequencies for the initial output probabilities at 0.... ..."
Cited by 2

Table 1: HMM Baseline

in Loosely coupled HMMs for ASR
by H. J. Nock 2000
"... In PAGE 3: ... Observa- tions comprise cepstra from two subbands 0-2 and 2-8kHz, with cepstral truncation (7,6), yielding a 39-d combined observation vector a20 a21 . Table1 gives baseline percentage correct (%C) per- formance of standard HMMs. Table 2 examines coupling through the transition matrices, using single Gaussian output distribu- tions.... ..."
Cited by 13

Table 2: Prepared HMM

in Application of Variational Bayesian Approach to Speech Recognition
by Shinji Watanabe, Yasuhiro Minami, Atsushi Nakamura, Naonori Ueda 2003
Cited by 9

Table 2: Prepared HMM

in unknown title
by unknown authors 2003
Cited by 9

Table 2: Prepared HMM

in Application of variational Bayesian approach to speech recognition
by Shinji Watanabe, Yasuhiro Minami, Atsushi Nakamura, Naonori Ueda 2003
Cited by 9

Table 6. HMM EBS

in Segmentation of Continuous Speech Using Acoustic-Phonetic Parameters and Statistical Learning
by Amit Juneja, Carol Espy-wilson 2002
Cited by 4

Table 2. Prepared HMM

in Application Of Variational Bayesian Estimation And Clustering To Acoustic Model Adaptation
by Shinji Watanabe, Yasuhiro Minami, Atsushi Nakamura, Naonori Ueda 2003
Cited by 3

Table 5.2 Segmentation rates of the HMM recognizer and the refinement procedure for all phonetic category at the ending boundaries of syllables. lt;=10 ms lt;=20 m lt;=30 ms gt;50 ms Phonetic category

in Automatic Segmentation and Labeling for Mandarin Chinese Speech Corpus for Concatenation-based TTS
by Cheng-yuan Lin, Jyh-shing Roger Jang, Kuan-ting Chen 2005
Cited by 2

Table 3: Comparison of HMM, HMM/poly 2 (2nd degree), and HMM/poly 3 (3rd degree). HMM HMM/poly 2 HMM/poly 3

in Using Polynomial Networks For Speech Recognition
by W. M. Campbell, C. C. Broun
"... In PAGE 8: ... A compar- ison of the standard approach and the new method is shown in Table 3. From Table3 , we can see that the HMM/polynomial combination performs well for all menus, but is not dramatically better than the standard Baum-Welch training for quadratic polynomials. We were somewhat disappointed by this result, since for the case of speaker verification, the new method of training has definite ad- vantages in accuracy and computation, see [2, 4].... In PAGE 8: ... We experimented using the log scoring technique instead of the sum technique shown in (6). When the probability was not allowed to go below BCBMBCBD, we obtained similar results to those shown in Table3 . Experimenting with choices other than BCBMBCBD did not drastically change the results.... ..."
Cited by 1

Table 2: Recognition error rates (%) for the trajectory HMM and the standard HMM.

in Trajectory Modeling based on HMMs with the Explicit Relationship Between Static and Dynamic Features
by Keiichi Tokuda, Heiga Zen, Tadashi Kitamura
"... In PAGE 3: ... This means that the trajectory HMM has the capability to capture the coarticulation effects naturally. Table2 shows the phoneme recognition error rates for the trajectory HMM and the standard HMM. The trajectory HMM achieves a relative error reduction of 7% over the standard... ..."
Results 1 - 10 of 1,276