Browse Prior Art Database

Offset Tracking in RPE/LTP Coder

IP.com Disclosure Number: IPCOM000035232D
Original Publication Date: 1989-Jun-01
Included in the Prior Art Database: 2005-Jan-28
Document File: 2 page(s) / 50K

Publishing Venue

IBM

Related People

Galand, C: AUTHOR [+2]

Abstract

At the analysis side of a RPE/LTP speech coder, the speech signal s(n) is analyzed by short term prediction, and filtered by the optimal inverse filter A(z). The residual signal r(n) is then analyzed by long- term prediction (LTP), and the error signal x(n) is modelized by regular pulse excited (RPE) coding. The LTP loop allows producing the estimated residual signal x''(n) to be subtracted from the input residual signal r(n) to give the error signal x(n). (Image Omitted) Referring to the figures, transmitted information consists in a set of parcor coefficients K(i) describing the vocal tract, a set of LTP coefficients (b,M), and the grid numbers j associated with the quantized sub-sampled sequences xj'(n).

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 59% of the total text.

Page 1 of 2

Offset Tracking in RPE/LTP Coder

At the analysis side of a RPE/LTP speech coder, the speech signal s(n) is analyzed by short term prediction, and filtered by the optimal inverse filter A(z). The residual signal r(n) is then analyzed by long- term prediction (LTP), and the error signal x(n) is modelized by regular pulse excited (RPE) coding. The LTP loop allows producing the estimated residual signal x''(n) to be subtracted from the input residual signal r(n) to give the error signal x(n).

(Image Omitted)

Referring to the figures, transmitted information consists in a set of parcor coefficients K(i) describing the vocal tract, a set of LTP coefficients (b,M), and the grid numbers j associated with the quantized sub-sampled sequences xj'(n). The decoded error signal is then upsampled, and forwarded to the LTP filter, the result of which is used to excite the short term filter 1/A(z). The LTP filter and the A(z) filter are updated, respectively, every 5 and 20 ms in order to take into account the time variations of the speech signal.

The RPE/LTP coder can achieve near toll quality compression of speech at medium bit rates, but presents the default to create audible tones if the signal to be coded presents an important continuous component. For example, this component (or offset) can be produced by an A/D converter. In the RPE/LTP coder, the high-frequency components are reconstructed by folding of the base- band of the quantized error signal. As a result, if the spe...