Showing papers on "Viseme published in 1988"

An investigation of young infants' perceptual representations of speech sounds.

Abstract: The present study examined the ability of newborns and 2-month-olds to detect phonetic differences between syllables By relying on the modified high-amplitude sucking procedure, which did not permit the infants to use a simple same-different response, the present experiments tapped the perceptual representations of the speech sounds Infants as young as a few days old displayed some capacity to represent differences in a set of syllables varying in their phonetic composition, although there was no convincing evidence that their representations were structured in terms of phonetic segments Finally, evidence of developmental changes in speech processing were noted for the first time with infants in this age range The change noted was a tendency from global toward more specific representations on the part of the older infants

Prosody and speech recognition

Abstract: The best ebooks about Prosody And Speech Recognition that you can get for free here by download this Prosody And Speech Recognition and save to your desktop. This ebooks is under topic such as prosody dependent speech recognition on radio news book reviews: prosody and speech recognition aclweb prosody modeling for automatic speech recognition and prosodic and accentual information for automatic speech modeling the prosody of hidden events for improved word prosody modeling for automatic speech understanding: an prosody in speech recognition ida which words are hard to recognize? prosodic, lexical, and prosody recognition in male infant-directed speech prosody-enriched lattices for improved syllable recognition using prosody for the improvement of automatic speech prosody as a conditioning variable in speech recognition prosody dependent speech recognition with explicit using prosody to improve automatic speech recognition prosody for mandarin speech recognition: a comparative use of prosodic features for speech recognition a prosody-only decision-tree model for disfluency detection recognition of prosodic factors and detection of landmarks towards using prosody in speech recognition/understanding prosodic parsing for swedish speech recognition prosody dependent speech recognition with explicit direct modeling of prosody: an overview of applications in recognition and understanding of prosody speech recognition university of maryland modeling prosodic dynamics for speaker recognition cnbc automatic detection of prosody phrase boundaries for text predicting automatic speech recognition performance using two methods for assessing oral reading prosody prosody unsupervised adaptation sail asa speech prosody pal aging and speech prosody illinois speech and language a study on prosody analysis ijcer prosody modeling in concept-to-speech generation the contributions of prosody and semantic context in how prosody improves word recognition modeling and recognition of phonetic and prosodic factors using prosody to improve mandarin automatic speech recognition implications of prosody modeling for prosody recognition the limits of speech recognition university of maryland prosody and focus in speech to infants and adults the prosody-voice screening profile (pvsp): psychometric applications 5: speech recognition theme 1 speech importance of prosodic features in language identification a factored language model for prosody dependent speech modeling and recognition of phonetic and prosodic factors improvement of speech summarization using prosodic information

Large-vocabulary speech recognition: a system for the Italian language

Abstract: We describe a research project in automatic speech recognition which has led to the development of an experimental large-vocabulary real-time recognizer for Italian, and show how the maximum-likelihood techniques which had been employed in the development of prototype recognizers for English can be tailored to a language with substantially different characteristics.

A new protocol for assessing viseme perception in sentence context: the lipreading discrimination test.

Abstract: Forty-six hearing-impaired young adults were tested with a newly developed instrument that requires a discrimination response to assess viseme perception as a component of lipreading performance. Stimuli were videotaped sentences that differed on half of the trials from a captioned target sentence by one viseme embedded in the middle of the sentence. Discrimination within six visual categories was tested: gross syllable pattern, consonant articulation--lips, consonant articulation--tongue, vowel articulation--extreme lip shapes, vowel articulation--graded lip shapes, and vowel articulation--jaw movement. Test data were analyzed using an item response theory model. The results indicated that the test data conformed to the expectations of the Rasch model for person measurement. Relationships among subjects' test scores and communication characteristics also were examined. The data provide evidence that the test protocol, at this early stage of development, is useful for assessing at least one perceptual component of lipreading performance.

Recognition of speaker-dependent continuous speech with Keal-Nevezh

Abstract: A description of the speaker-dependent continuous speech understanding system KEAL-NEVEZH is given An unknown utterance is recognized by means of the following procedures: Acoustic analysis, phonetic segmentation and identification, word and sentence analysis This new system is an extension of the KEAL system, connected to ALOEMDA, an active chart parser modifying its strategy and linguistic capabilities

Speech recognition based on speech units

Abstract: In a classical quantization system, each vector is represented by the nearest centroid; two vectors belonging to the same class are then indistinguishable. In order to mitigate this situation, we take into account the two nearest neighbours and define a “belonging degree” calculated from the distances between the vector and the two centroids. In the case of speaker independent speech recognition system, this “fuzzy” quantization gives better results.