The application of a functional perceptual model of speech to variable-rate LPC systems

Viswanathan, R.; Makhoul, John; Wicke, R. W.

doi:10.1109/icassp.1977.1170369

Cited by 7 publications

(2 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Thus, it would be more realistic if the coefficient matrix would adapt 1 Here, the v n sequence is equivalent to the delta coefficients used in many ASR systems. 2 In this case, vn forms the delta-delta, or "acceleration," coefficients used in more sophisticated ASR systems. 3 In ASR applications, this training can be achieved by tying the model's coefficient matrices to individual states within a hidden Markov model (HMM), to give a linear predictive HMM [16].…”

Section: Nonstationary Modelsmentioning

confidence: 99%

“…Thus, if a link has been identified between element of the th observation vector, , and element of , then it is assumed that where . In other words (2) If more than one link terminates in a single element of , the values estimated for each link are averaged. The identification of links is performed by a similar method to that used to create the spectrographic flow described previously: the only difference is that additional constraints are applied so that no link is allowed to imply that any feature will move beyond the end of the vector, , or that any element of will be left without a link to at least one element in .…”

Section: A Flow-based Predictionmentioning

confidence: 99%

See 1 more Smart Citation

Multistep coding of speech parameters for compression

Baghai-Ravary¹,

Beet

1998

IEEE Trans. Speech Audio Process.

View full text Add to dashboard Cite

This paper presents specific new techniques for coding of speech representations and a new general approach to coding for compression that directly utilizes the multidimensional nature of the input data.Many methods of speech analysis yield a two-dimensional (2-D) pattern, with time as one of the dimensions. Various such speech representations, and power spectrum sequences in particular, are shown here to be amenable to 2-D compression using specific models which take account of a large part of their structure in both dimensions. Newly developed techniques, multistep adaptive flux interpolation (MAFI) and multistep flow-based prediction (MFBP) are presented. These are able to code power spectral density (PSD) sequences of speech more completely and accurately than conventional methods. This is due to their ability to model nonstationary, but piecewise-continuous, signals, of which speech is a good example.Initially, MAFI and MFBP are applied in the time domain, then reapplied to the encoded data in the second dimension. This approach allows the coding algorithm to exploit redundancy in both dimensions, giving a significant improvement in the overall compression ratio. Furthermore, the compression may be reapplied several times. The data is further compressed with each application.

show abstract