Browse Prior Art Database

Variable-Rate Time-Domain Harmonic Scaling Speech Communication System

IP.com Disclosure Number: IPCOM000041727D
Original Publication Date: 1984-Mar-01
Included in the Prior Art Database: 2005-Feb-02
Document File: 4 page(s) / 45K

Publishing Venue

IBM

Related People

Irvin, DR: AUTHOR

Abstract

Variable-rate speech coders have been proposed for multiplexing onto a common communications channel. This produces a statistical averaging of resource demand peaks and valleys which effectively reduces the average channel rate required for each user. This, in turn, enables more data to be multiplexed into the speech communication channel or, alternatively, allows more speech users access to the channel. The present technique is an improvement that deals with a method of operating a time-domain harmonic scaling speech (TDHS) coder at a variable-demand assigned coding rate. Variable compression of a block of speech is achieved by assigning a variable compression rate based upon an algorithm described below. The degree of compression is based upon a signal-to-noise calculation.

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 28% of the total text.

Page 1 of 4

Variable-Rate Time-Domain Harmonic Scaling Speech Communication System

Variable-rate speech coders have been proposed for multiplexing onto a common communications channel. This produces a statistical averaging of resource demand peaks and valleys which effectively reduces the average channel rate required for each user. This, in turn, enables more data to be multiplexed into the speech communication channel or, alternatively, allows more speech users access to the channel. The present technique is an improvement that deals with a method of operating a time-domain harmonic scaling speech (TDHS) coder at a variable-demand assigned coding rate. Variable compression of a block of speech is achieved by assigning a variable compression rate based upon an algorithm described below. The degree of compression is based upon a signal-to-noise calculation. Fixed rate TDHS coders and decoders are well known in the art and have been described in the IEEE Transactions on Acoustics, Speech and Signal Processing ASSP-27, 121-133 (April, 1979) and in the Bell System Telephone Journal 60, 2107-2156 (November, 1981). A variable-rate TDHS coder was not described. Similarly, the specific assignment algorithm based on the signal-to-noise ratio measurements for assigning the compression rate is also believed to be new. The theoretical basis for the operation of the TDHS coder is given in the reference noted above. The operation is simply described as a combination of windowing and signal splicing. Fig. 1 shows a basic block diagram of such a system. On line 1, incoming analog speech signals are applied to an analog-to-digital (A/D) converter 2. The output on line 3 is digitally coded speech in accordance with the type of A/D converter employed. A pitch extraction circuit defines pitch periods in the speech digital pattern and it, in combination with the digital pattern, is applied to a compression circuit 5 to remove redundancy and achieve a decrease in the net bandwidth required for transmission. The compression algorithm supplies compression control on line 6 to a multiplexer 7 which also receives the digital stream after its exit from the pitch extraction circuit 4. Communication channel 8 is fed by the multiplexer 7 from several sources, although only one source is shown. A demultiplexer 9 at the receiving end of the communication line 8 decodes the pitch period in decoder 10 and causes the inverse of the compression to occur in expansion circuit 11. The digital-to-analog (D/A) conversion is then performed on the expanded signal resulting in the reconstruction of a representation of the original analog input on line 1. Fig. 2 schematically shows some of the details for a 2 to 1 compression that will be conducted in a compression circuit 5 operated by the input stream from an A/D converter 2 and a pitch extraction circuit 4, as shown in Fig. 1. In Fig. 2, the top line A shows a period of time in which a burst of signal energy is found by a TASI (Time...