Transform vector quantization for adaptive predictive coding

• US 5,487,086 A
• Filed: 09/13/1991
• Issued: 01/23/1996
• Est. Priority Date: 09/13/1991
• Status: Expired due to Term

• Current Assignee: Intelsat Global Service Corporation
• Original Assignee: COMSAT Corporation (Reorganized ISA SA)
• Inventors: Bhaskar, B. R. Udaya
• Primary Examiner(s): Chin, Stephen
• Assistant Examiner(s): VO, DON NGUYEN
• Application Number:
  US07/759,361
• Time in Patent Office: 1,593 Days
• Field of Search: 375/26, 375/27, 375/122, 381/31, 381/36, 381/41, 381/43, 381/29, 381/30, 358/133, 358/135, 358/136, 382/41, 382/43, 382/56, 348/394, 348/395, 348/403, 348/405, 348/414, 348/417, 348/422
• US Class Current: 375/243
• CPC Class Codes: G10L 19/0212 using orthogonal transformation
• CPC Class Codes: G10L 19/04 using predictive techniques
• CPC Class Codes: G10L 2019/0013 Codebook search algorithms

Abstract

Before transmitting signals to a receiver, the signals are subjected to adaptive prediction to generate a residual signal for transmission, and the residual signal is then transformed into frequency domain coefficients, the coefficients are grouped together to form vectors, and the vectors are then quantized.

Claims

1. An apparatus for processing digital information signals at a transmitter end of a communications system before said signals are transmitted to a receiver end, said apparatus comprising:

input means for receiving an input digital signal;

adaptive prediction means for performing adaptive prediction upon said input digital signal received by said input means; and

transform domain vector quantization means for transforming the output of said adaptive prediction means into frequency domain coefficients, grouping said coefficients into vectors and quantizing said vectors, wherein said transform domain vector quantization means further includes;

an input signal power spectrum estimation means for estimating the input signal power spectrum of said input digital signal; and

coefficient grouping means for grouping said coefficients into vectors in an adaptive manner based on results obtained from said input signal power spectrum estimation means.

2. An apparatus according to claim 1 wherein said transform domain vector quantization means further includes:

a means for making an average log magnitude synthesis response for each vector substantially equal to the average log magnitude synthesis response for all the frequency domain coefficients.

3. An apparatus for processing digital information signals at a transmitter end of a communications system before said signals are transmitted to a receiver end, said apparatus comprising:

input means for receiving an input digital signal;

adaptive prediction means for performing adaptive prediction upon said input digital signal received by said input means, said adaptive prediction means including both a short term predictor and a long term predictor; and

transform domain vector quantization means for transforming the output of said adaptive prediction means into frequency domain coefficients, grouping said coefficients into vectors, and quantizing said vectors, said transform domain vector quantization means includes quantizer means for quantizing the vectors by using a weighting function that takes into account auditory noise masking properties of the human ear as well as parameters obtained from said short and long term predictors.

4. An apparatus according to claim 3 wherein said quantizer means includes a codebook of possible codevectors.

5. An apparatus for processing digital information signals at a transmitter end of a communications system before said signals are transmitted to a receiver end, said apparatus comprising:

input means for receiving an input digital signal; and

transform domain vector quantization means for transforming the received input digital signals into frequency domain coefficients, grouping said coefficients into vectors and quantizing said vectors,

wherein said transform domain vector quantization means further includes a means for making an average log magnitude synthesis response for each vector substantially equal to the average log magnitude synthesis response for all the frequency domain coefficients.