Method and apparatus for coding audio signals based on perceptual model

• US 5,285,498 A
• Filed: 03/02/1992
• Issued: 02/08/1994
• Est. Priority Date: 03/02/1992
• Status: Expired due to Term

• Current Assignee: AT&T IPM Corp. (AT&T, Inc.)
• Original Assignee: AT&T, Inc.
• Inventors: Johnston, James D.
• Primary Examiner(s): Isen, Forester W.
• Application Number:
  US07/844,804
• Time in Patent Office: 708 Days
• Field of Search: 381/2, 381/94
• US Class Current: 381/2
• CPC Class Codes: H04B 1/665 using psychoacoustic properties of the ear, e.g. masking effect
• CPC Class Codes: H04H 20/88 Stereophonic broadcast systems multiplex system in general H04J

Abstract

Coding of high quality stereophonic audio signals is accomplished in a perceptual filterbank coder which exploits the interchannel redundancies and psychoacoustic. Using perceptual principles, switching between a normal and short window of input samples improve output signal quality for certain input signals, particularly those having a rapid attack. Switching is also accomplished between coding of left and right channels and so- called sum and difference channels in response to particular signal conditions. A number of new perceptually based techniques, including improved threshold determinations, result in high quality.

Claims

1. A method for coding an input set of stereophonic audio signals comprising respective left channel and right channel sets of signals, said method comprising

forming for each of said left channel and right channel sets of signals, a set of first signals representing the frequency content of said input set, said set of first signals comprising signals representing amplitude and phase information for each of a plurality of frequency bands,

forming sets of sum and difference channel signals corresponding, respectively, to the sum of, and difference between, corresponding ones of said set of first signals for said left channel and said set of first signals for said right channel,

determining a randomness metric for each of said plurality of frequency bands for each of said sets of first signals,

based on the distribution of power with frequency for each of said sets of first signals and on said randomness metrics for each of said sets of first signals, forming a tonality function as a function of frequency, and

based on said randomness metric determining a first perceptual threshold for each of said left channel and right channel sets of signals and selecting between (i) said sum and difference channel signals, or (ii) said left channel and right channel signals, based on the determined threshold values for each of said plurality of frequency bands.

2. The method of claim 1, comprising selecting said sum and difference channel signals for those frequency bands for which the first signals differ by less than a predetermined amount.

3. The method of claim 2, wherein said predetermined amount is substantially equal to 2 dB.

4. The method of claim 2, wherein L sub i is the ith first signal for said left channel and R sub i is the ith first signal for said right channel, and said sum channel signal is given by M sub i=(L sub i+R sub i)/2.