Browse Prior Art Database

TDHS Preprocessing Compression Selectively Applied to Sub-Band Speech Coders

IP.com Disclosure Number: IPCOM000041955D
Original Publication Date: 1984-Mar-01
Included in the Prior Art Database: 2005-Feb-03
Document File: 3 page(s) / 81K

Publishing Venue

IBM

Related People

Irvin, DR: AUTHOR

Abstract

Multiple user voice systems often utilize statistical signal patterns in conventional speech and identify blanks or gaps of sounds in which other conversations can be compressed so that the entire demand can be compressed into a lesser number of communication channels to afford bandwidth or channels to other potential users. Stochastically, the number of simultaneously active users will, on some occasions, exceed the number of available channels. On such occasions, some users are "frozen out" and system performance will be degraded. This disclosure describes a multi-user sub-band coding system of a sort known in the art but with improved channel compression provided by selective application of TDHS (time domain harmonic scaling) precompression.

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 41% of the total text.

Page 1 of 3

TDHS Preprocessing Compression Selectively Applied to Sub-Band Speech Coders

Multiple user voice systems often utilize statistical signal patterns in conventional speech and identify blanks or gaps of sounds in which other conversations can be compressed so that the entire demand can be compressed into a lesser number of communication channels to afford bandwidth or channels to other potential users. Stochastically, the number of simultaneously active users will, on some occasions, exceed the number of available channels. On such occasions, some users are "frozen out" and system performance will be degraded. This disclosure describes a multi-user sub-band coding system of a sort known in the art but with improved channel compression provided by selective application of TDHS (time domain harmonic scaling) precompression. Sub-band multi-user coding systems alleviate a great deal of system degradation during periods of freeze-out. However, these coders either deny access to some users, which introduces clipping in leading speech segments, or steal channel assignment bits from all the active coders, which reduces the overall average coding rate and adversely affects the subjective performance of the system. In low-bit-rate speech coders, this aspect is especially severe. This disclosure minimizes the perceived performance degradation and maintains integrity in the leading speech segments. A small group of TDHS coders are shared across a large group of sub-band coders (SBCs) on a statistical basis. During overload conditions, signals from the users determined to contain patterns most amenable to further bit rate reduction via the TDHS preprocessing technique are identified by an algorithm described below. For the duration of the coding frame, these sources are coded by a combined TDHS and SBC operation to operate at one- half the bit rate for the remaining unassisted SBC coders. The TDHS preprocessing for selected signal sources provides an additional compression in the range of 1.5 to 1 to 2 to 1 and introduces only a slight degradation in subjective performance. An optimum TDHS compression is achieved on highly regular and repetitious waveforms which exhibit a compact spectrum. During normal operation of sub-band coders, frame by frame estimates of the energy content are generated as a byproduct of the sub-band coder operation. This information is utilized in the present disclosure to form an evaluation of the signal spectral compactness. Those incoming signals having a highly concentrated energy spectra are assigned a high figure of merit and are those which are first subjected to the TDHS preprocessing to free up additional coding bits for other users. The operations of sub-band coders and TDHS coders are well known and have been described together in an article by D. Malah, et al., entitled "Performance of Transform and Sub-band Coding Systems Combined with Harmonic Scaling of Speech," appearing in IEEE Transactions on Acoustics, ...