Showing papers in "Speech Communication in 1988"

TL;DR: The articulatory-acoustic relation is simplified, the quantal nature of speech is confirmed, phonetic universals and phonetic systems are considered from a new point of view, formant transitions are explained and normalized, and an easy to control 9-parameter model proposed.

TL;DR: In this article, the authors describe sphinx, the world's first accurate large-vocabulary speaker-independent continuous speech recognition system, and compare its performance against similar systems, and account for its high accuracy.

TL;DR: Improvements to the SELP algorithm are described which result in better speech quality and higher computational efficiency, and a new recursive algorithm which performs a very fast search through the adaptive codebook.

TL;DR: Different methods for the second task are reviewed, emphasizing the advantages and disadvantages of the linear predictive (LPC) diphone approach.

TL;DR: The choice of the parameter values used by the Multi-Layer Perceptron is discussed and experimental results are quoted to show how the choice of these parameter values influences the performance of the MLP.

TL;DR: The conclusions suggest the use of time-compressing preprocessing techniques and the application of suboptimal DTW procedures as the most likely causes of the TI dissatisfaction rates reported elsewere.

TL;DR: It is suggested that the acoustic context provides information about the formant frequencies of the talker's vowels with which a vowel space can be constructed that serves as a reference frame for the identification of the vowels in the test words.

TL;DR: Data on stop sequences in French which supports the corpoductionist's point of view is presented and the articulatory pattern observed by electropalatography cannot be interpreted simply as the concatenation of assimilated segments.

TL;DR: Both objective and subjective results confirm the high level of performances obtained by the 16 kbit/s CELP coder in different realistic transmission conditions as transmission with errors and ambient noise.

TL;DR: A 16 kbit/s speech codec with low complexity and low signal delay is presented which is a special version of the Regular-Pulse Excitation LPC approach (RPE-LPC).

TL;DR: The cross-correlation coefficient was used to investigate LTS residual intra-speaker variability both in inter- and intra-text conditions, and significant subject-dependent differences have been revealed in both conditions.

TL;DR: The Multipulse excitation with long term prediction (MPE/LTP) algorithm and details on the implementation (constants, quantizing tables) respectively in the analysis part, the error protection/correction, and the synthesis part of the codec are reported.

TL;DR: It is shown that, by increasing the number of tokens included in dictionaries with multiply represented words, a simultaneous reduction can be achieved in both the error-rate and thenumber of distance computations required.

TL;DR: The results show that a complexity reduction of about 73% can be achieved by using the two pass approach with respect to the direct approach, while the recognition accuracy remains comparable.

TL;DR: The main result reported in the paper is that the performances of the two schemes are almost equivalent although their structure is very different.

TL;DR: A Regular Pulse Excitation/Long-Term Prediction LPC (RPE-LTP) coding algorithm has been selected as the basis for the standard for the Pan-European cellular system.

TL;DR: In this article, the authors describe subjective testing methodologies adopted to select suitable candidate codecs capable of being used in the proposed Pan-European cellular digital mobile radio (DMR) system.

TL;DR: An overview of a system for phoneme-based large-vocabulary continuous-speech recognition that provides the speaker dependent recognition component in the speech understanding system spicos that is designed to recognize and understand database queries spoken in natural German language.

TL;DR: An 8 kbit/s simulation is presented, using hard switching between harmonic coding and ATC to discuss the state of the art in analysis-synthesis methods and their application to coding.

TL;DR: A Markov-modelling Spellmode recognizer is described which uses LPC-VQ as a front-end for analog to digital conversion and data compression and it suffers from high computational cost.

TL;DR: A “coding gap” of roughly 32-2.4 kbit/s is shown to actually define “medium-rate” speech coding, and the fundamental approaches trying to close the gap are exposed.

TL;DR: All the 1268 syllables in Standard Chinese have been synthesized by this system, which produces a sound quality close to that of natural speech with respect to both intelligibility and naturalness.

TL;DR: Modified versions of Edited and Condensed Nearest Neighbor Rules are applied to speaker-independent isolated word recognition to select the word templates, as opposed to the clustering techniques.

TL;DR: A paradigm for automatic speech recognition using networks of actions performing variable depth analysis produces descriptions of speech properties that are related to speech units through Markov models representing system performance.

TL;DR: The 16 kbit/s speech codec developed by British Telecom Research Laboratories and selected as the UK candidate to the GSM Pan-European study on digital cellular land mobile radio offers several important features including low delay, low computational complexity and a good tolerance to transmission errors.

TL;DR: A polynomial analysis of the vocal tract transfer function was done to obtain new practical models for the Higher Pole Correction (HPC), which can be used in analog as well as digital all-pole realizations to form a new type of pole-zero model for speech production.

TL;DR: Elsag’s Large Vocabulary Isolated Word Recognition system DSPELL makes use of a diphone-based speech model and an extremely efficient word decoding algorithm, and is implemented on Elsag's multiprocessor EMMA-2 1.

TL;DR: A verification algorithm and a word spotting technique both based on HMM will be discussed and some preliminary results for the matching procedures are given.

TL;DR: The nature du signal de parole conduit a envisager un traitement different for les zones transitoires and les zones stables as mentioned in this paper, and le but est de delivrer un treillis d'hypotheses quasi phonetiques le plus precis possible.