Wu, S.-L., Shire, M. L., Greenberg, S., and Morgan, N. Integrating Syllable Boundary Information into Speech Recognition. In Proceedings of ICASSP '97 (Munich, Germany, April 1997), pp. 987-990.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....structure that includes units larger than the syllable implies domains of co articulations beyond the syllable, syllable level structure must at least be a factor in acoustic variability. Acoustic features that are closely related to syllables have been used in speech recognizers. Wu et al. [128] detected syllable boundaries using a neural network, and scored phone sequences at syllable boundaries using a syllable grammar. A potential problem with this approach is that detection of the syllable boundaries is unlikely to be robust in spontaneous speech. This approach has the general ....

Su-Lin Wu, Michael L. Shire, Steven Greenberg, and Nelson Morgan. Integrating syllable boundary information into speech recognition. Proc. Int'l Conf. on Acoustic, Speech and Signal Processing, 2:987-990, 1997.

....following methods attempt to accurately detect the syllable boundaries, given an input continuous speech signal. They accomplish this task by some form of perceptually oriented feature extraction and their consequent post processing. 2.1. First method This method is based on the work of Wu et al. [4] where they present a set of features based on the analysis of energy trajectories in critical band channels. Large values in this representation correspond to patterns of synchronized rises and falls in subband energy regions where syllable onset characteristics occur. These syllable length ....

S. Wu, M. L. Shire, S. Greenberg, N. Morgan, Integrating Syllable Boundary Information into Speech Recognition, in Proceedings ICASSP 97, Munich, Germany, 1997.

....models with 18135 gaussians. 4.2 New features In (Fissore et al. 1997) word temporal structures are proposed in a two dimensional cepstral feature set. Syllable boundary information is integrated into the decoding process with a slight improvement in number recognition with the OGI corpus (Wu et al. 1997). Modeling of phonetic segments with multi resolution spectral time correlates (cepstral features) is proposed in (Vaseghi et al. 1997) Phonetic features are derived as trajectories of log spectral energies at the output of a filter bank. Multi resolution cepstrum is also considered using 6 ....

Wu S.L., Shire M., Greenberg S. and Morgan N. (1997) Integrating syllable boundary information into speech recognition.

....somewhat constraining a phoneme based system to align on syllables, but also allowing to make use of microprosodic features like syllabic stress. The method was shown to yield significant performance improvement. As an alternative approach to introducing syllabic constraints, it was recently shown [10] that acoustically derived syllable onsets can improve speech recognition performance. This further motivates research towards efficient use of long time regions (covering 200 ms or more) and syllable level information. This preliminary work will now be extended in several directions. Experiments ....

S. Wu, M. Shire, S. Greenberg, and N. Morgan, "Integrating syllable boundary information into speech recognition," in Proc. IEEE Internat. Conf. Acoust. Speech and Signal Process., (Munich), pp. 987--990, Apr. 1997.

....Syllable Boundary Information This section reports experiments aimed at improving recognition accuracy by incorporating syllable boundary information during search. Previous research on detecting syllable boundaries and using this information to improve recognition accuracy has been reported [7, 8]. In this work we use the method of Wu et al. [7] 3.1. Detecting Syllable Boundaries The broadcast news training data does not include syllable boundary or phonetic alignment information. An automatic procedure for determining syllable boundaries is therefore required. The method used in this ....

....experiments aimed at improving recognition accuracy by incorporating syllable boundary information during search. Previous research on detecting syllable boundaries and using this information to improve recognition accuracy has been reported [7, 8] In this work we use the method of Wu et al. [7]. 3.1. Detecting Syllable Boundaries The broadcast news training data does not include syllable boundary or phonetic alignment information. An automatic procedure for determining syllable boundaries is therefore required. The method used in this work is based on deriving syllable boundaries from ....

[Article contains additional citation context not shown here]

S-L. Wu, M.L. Shire, S. Greenberg, and N.Morgan. Integrating Syllable Boundary Information into Speech Recognition. International Conference on Acoustics, Speech, and Signal Processing, 2:987--990, April 1997. Berlin.

....SYLLABLE BOUNDARY INFORMATION This section reports experiments aimed at improving recognition accuracy by incorporating syllable boundary information during search. Previous research on detecting syllable boundaries and using this information to improve recognition accuracy has been reported [18, 6]. In this work we use the method of Wu et al. [18] 5.1. Detecting Syllable Boundaries The broadcast news training data does not include syllable boundary or phonetic alignment information. An automatic procedure for determining syllable boundaries is therefore required. The method used in this ....

....experiments aimed at improving recognition accuracy by incorporating syllable boundary information during search. Previous research on detecting syllable boundaries and using this information to improve recognition accuracy has been reported [18, 6] In this work we use the method of Wu et al. [18]. 5.1. Detecting Syllable Boundaries The broadcast news training data does not include syllable boundary or phonetic alignment information. An automatic procedure for determining syllable boundaries is therefore required. The method used in this work is based on deriving syllable boundaries from ....

[Article contains additional citation context not shown here]

S-L. Wu, M.L. Shire, S. Greenberg, and N.Morgan. Integrating Syllable Boundary Information into Speech Recognition. International Conference on Acoustics, Speech, and Signal Processing, 2:987--990, April 1997. Berlin.

....tested on much larger corpora of spontaneous speech. Some encouraging news is that many of the false alarms for unknown words occur where there is a recognition error, so that the model is actually applicable in those cases. 1.2. 2 Incorporating Syllable Boundaries in Speech Recognition Wu et al. [14] use syllabic boundaries to improve speech recognition performance. These boundaries are derived from two different sources. One is from the acoustic signal, and the other is from the word: class: morpheme: grapheme: adjective pref root suf o u t s t a n d i n g Figure 1 1: Part of a ....

Wu, S., M. Shire, S. Greenberg, and N. Morgan, "Integrating Syllable Boundary Information into Speech Recognition," Proceedings of ICASSP-97, Munich, Germany, April 1997, pp. 987-990.

....cepstrum) Neural networks (NNs) that show a high performance for many classification tasks are suitable for estimating accurate boundaries. There have recently been several reports that boundary information obtained from feed forward multilayer perceptrons (MLP) improves recognition performance [4][5] In this paper, we propose a segment (e.g. phoneme) boundary estimation method based on bi directional recurrent neural networks (BRNNs) A BRNN can be trained without the limitation of using a fixed size input window, and it gave better classification performance than a regular RNN on test ....

S.-L. Wu, M. Shire, S. Greenberg and N. Morgan : "Integrating syllable boundary information into speech recognition," Proc. ICASSP-97, pp. 987--990, 1997.

....provides a basis for the syllable to function as a fundamental grouping element for speech, reflecting the acoustic realization of the articulatory gesture. Given the potential utility of syllabic information, how practical is it to segment the speech signal into such units automatically Shire [44] has shown that it is possible to accurately determine the syllabic onsets about 85 90 of the time, using relatively simple signal processing techniques that look for patterns of energy over contiguous spectral channels. And Kingsbury [20] has shown that a representation of the speech signal ....

Wu, S-L., Shire, M., Greenberg, S. and Morgan, N. (1997) Integrating syllable boundary information into speech recognition, in ICASSP-97, IEEE International Conference on Acoustics, Speech and Signal Processing, Munich.

No context found.

Wu, S.-L., Shire, M. L., Greenberg, S., and Morgan, N. Integrating Syllable Boundary Information into Speech Recognition. In Proceedings of ICASSP '97 (Munich, Germany, April 1997), pp. 987-990.

No context found.

S.-L. Wu, M. L. Shire, S. Greenberg, and N. Morgan, "Integrating syllable boundary information into speech recognition," in Proc. IEEE Intern. Conf. Acoust., Speech, Signal Process. (ICASSP), vol. 2, 1997, pp. 987--990.

No context found.

Wu, S.L.; Shire, M.; Greenberg, S.; Morgan, N., 1997. Integrating syllable boundary information into speech recognition. IEEE International Conference on Acoustics, Speech and Signal Processing. Munich, 987-990.

No context found.

S.-L. Wu, M. Shire, S. Greenberg, and N. Morgan. Integrating syllable boundary information into speech recognition. In Proceedings of ICASSP, pages 987-990, 1997.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC