Computationally efficient adaptive bit allocation for encoding method and apparatus with allowance for decoder spectral distortions

• US 5,623,577 A
• Filed: 01/28/1994
• Issued: 04/22/1997
• Est. Priority Date: 07/16/1993
• Status: Expired due to Term

• Current Assignee: Dolby Laboratories Licensing Corporation (Dolby Laboratories Incorporated)
• Original Assignee: Dolby Laboratories Licensing Corporation (Dolby Laboratories Incorporated)
• Inventors: Fielder, Louis D.
• Primary Examiner(s): COLLINS, ALPHONSO
• Assistant Examiner(s): Sartori, Michael A.
• Application Number:
  US08/190,655
• Time in Patent Office: 1,180 Days
• Field of Search: 395/2, 395/2.1, 395/2.3-2.39, 381/36-40, 381/46, 381/47
• US Class Current: 704/200.1
• CPC Class Codes: H04B 1/665 using psychoacoustic properties of the ear, e.g. masking effect
• CPC Class Codes: H04B 1/667 using a division in frequency subbands for TV signals H04N19/63

Abstract

The invention relates in general to low bit-rate encoding and decoding of information such as audio information. More particularly, the invention relates to computationally efficient adaptive bit allocation and quantization of encoded information useful in high-quality low bit-rate coding systems.

In audio applications, a digital split-band encoder splits an input signal into frequency subband signals having bandwidths commensurate with the critical bandwidths of the human auditory system, quantizes the subband signals according to values established by an allocation function, and assembles the quantized subband signals into an encoded signal. The allocation function establishes allocation values in accordance with psychoacoustic principles with allowance for decoding synthesis filter bank spectral distortions.

In one embodiment, an allocation function establishes allocation values using a psychoacoustic masking threshold generated by estimating the power spectral density (PSD) of the input signal, generating an excitation pattern by applying a basilar-membrane spreading function to the PSD, adjusting the excitation pattern by an amount equal to a sensitivity function which specifies a signal-to-noise ratio (SNR) sufficient to achieve psychoacoustic masking, comparing the level of the adjusted pattern to the threshold of hearing and generating the psychoacoustic masking threshold which is equal to the larger of the two. An allocation function may allow for decoder synthesis filter bank spectral distortions in any of a number of ways such as by adapting the sensitivity function.

Claims

1. An encoding method for constructing an encoded representation of an input signal for decoding by a decoding method that comprises applying a plurality of synthesis filters, said encoding method comprising

generating subband signals by applying a plurality of analysis filters to said input signal;

producing, in response to said subband signals, encoded information having lower informational requirements than said subband signals, wherein said lower information requirements are achieved using a masking threshold established according to psycho-perceptual principles and by allowing for spectral distortions introduced by said decoding method, said allowing for spectral distortions being responsive to rate of rolloff of said masking threshold and rate of rolloff of frequency response of said plurality of synthesis filters; and

assembling said encoded information into said encoded representation.

2. An encoding method according to claim 1 wherein said plurality of analysis filters is implemented by one or more digital transforms.

3. An encoding method according to claim 1 wherein said allowing for said spectral distortions comprises

establishing allocation values in accordance with said psycho-perceptual principles;

establishing a noise spectrum in accordance with said allocation values;

estimating perceptual effects of decoding spectral distortion by generating a decoding distortion spectrum in response to said noise spectrum; and

adjusting said allocation values in response to said estimated perceptual effects of said decoding spectral distortion.

4. An encoding method according to claim 3 wherein said estimating perceptual effects comprises

generating said decoding distortion spectrum by convolving said noise spectrum with a respective frequency response of a respective one of said plurality of synthesis filters;

and

estimating said perceptual effects by establishing which if any portions of said decoding distortion spectrum exceeds said masking threshold.

5. An encoding method according to claim 4 wherein said adjusting comprises selecting noise spectrum components which are large contributors to portions of said decoding distortion spectrum exceeding said masking threshold, and adjusting allocation values of encoded information corresponding to said selected noise spectrum components.

6. An encoding method according to claim 5 wherein said selecting noise spectrum components comprises

weighting said noise spectrum with a frequency-reversed representation of said respective frequency response centered about each subband in which said decoding distortion spectrum exceeds said masking threshold;

generating a weighted noise spectrum by summing the results of said weighting for each noise spectrum component; and

selecting noise spectrum components corresponding to large weighted noise spectrum components.

7. An encoding method according to claim 1 wherein said allowing for spectral distortions comprises

estimating a spectral envelope of said input signal;

comparing said spectral envelope with said rate of rolloff of frequency response;

identifying one or more portions of said spectral envelope which change at a rate substantially equal to or greater than said rate of rolloff of frequency response; and

increasing information requirements of selected encoded information corresponding to said one or more portions of spectral envelope.

8. An encoding method according to claim 7 wherein information requirements of said selected encoded information is increased by reducing one or more portions of said masking threshold.

9. An encoding method according to claim 7 wherein information requirements of encoded information at frequencies above said one or more portions of spectral envelope are increased by an amount that decreases with increasing frequency.

10. An apparatus for constructing an encoded representation of an input signal for decoding by a decoder that comprises applying a plurality of synthesis filters, said apparatus comprising

means for generating subband signals by applying a plurality of analysis filters to said input signal;

means for producing, in response to said subband signals, encoded information having lower informational requirements than said subband signals, said means for producing encoded information comprising

means for achieving said lower information requirements using a masking threshold established according to psycho-perceptual principles; and

means, responsive to rate of rolloff of said masking threshold and rate of rolloff of frequency response of said plurality of synthesis filters,

for allowing for spectral distortions introduced by said decoder; and

means for assembling said encoded information into said encoded representation.

11. An apparatus according to claim 10 wherein said plurality of analysis filters is implemented by one or more digital transforms.

12. An apparatus according to claim 10 wherein said means for allowing for said spectral distortions comprises

means for establishing allocation values in accordance with said psycho-perceptual principles;

means for establishing a noise spectrum in accordance with said allocation values;

means for estimating perceptual effects of decoding spectral distortion by generating a decoding distortion spectrum in response to said noise spectrum; and

means for adjusting said allocation values in response to said estimated perceptual effects of said decoding spectral distortion.

13. An apparatus according to claim 12 wherein said means for estimating perceptual effects comprises

means for generating said decoding distortion spectrum by convolving said noise spectrum with a respective frequency response of a respective one of said plurality of synthesis filters;

and

means for estimating said perceptual effects by establishing which if any portions of said decoding distortion spectrum exceeds said masking threshold.

14. An apparatus according to claim 13 wherein said means for adjusting comprises

means for selecting noise spectrum components which are large contributors to portions of said decoding distortion spectrum exceeding said masking threshold;

and wherein allocation values of encoded information corresponding to said selected noise spectrum components are adjusted.

15. An apparatus according to claim 14 wherein said means for selecting noise spectrum components comprises

means for weighting said noise spectrum with a frequency-reversed representation of said respective frequency response centered about each subband in which said decoding distortion spectrum exceeds said masking threshold;

means for generating a weighted noise spectrum by summing the results of said weighting for each noise spectrum component; and

means for selecting noise spectrum components corresponding to large weighted noise spectrum components.

16. An apparatus according to claim 10 wherein said means for allowing for spectral distortions comprises

means for estimating a spectral envelope of said input signal;

means for comparing said spectral envelope with said rate of rolloff of frequency response;

means for identifying one or more portions of said spectral envelope which change at a rate substantially equal to or greater than said rate of rolloff of frequency response; and

means for increasing information requirements of selected encoded information corresponding to said portions of spectral envelope.

17. An apparatus according to claim 16 wherein information requirements of said selected encoded information is increased by reducing one or more portions of said masking threshold.

18. An apparatus according to claim 16 wherein information requirements of encoded information at frequencies above said one or more portions of spectral envelope are increase by an amount that decreases with increasing frequency.

19. An apparatus for decoding an encoded representation of an input signal, said decoder comprising

means for extracting quantized subband information from said encoded representation;

allocation means for establishing allocation values by using a masking threshold established according to psycho-perceptual principles and by allowing for spectral distortions introduced by a plurality of synthesis filters, said allowing for spectral distortions being responsive to rate of rolloff of said masking threshold and rate of rolloff of frequency response of said plurality of synthesis filters;

means for generating dequantized subband information by dequantizing said quantized subband information in accordance with said allocation values; and

means for generating an output signal by applying said plurality of synthesis filters to said dequantized subband information.

20. An apparatus according to claim 19 wherein said allocation means comprises

means for establishing allocation values in accordance with said psycho-perceptual principles;

means for establishing a noise spectrum in accordance with said allocation values;

means for estimating perceptual effects of decoding spectral distortion by generating a decoding distortion spectrum in response to said noise spectrum; and

means for adjusting said allocation values in response to said estimated perceptual effects of said decoding spectral distortion.

21. An apparatus according to claim 20 wherein said means for estimating perceptual effects comprises

means for generating said decoding distortion spectrum by convolving said noise spectrum with a respective frequency response of a respective one of said plurality of synthesis filters;

and

means for estimating said perceptual effects by establishing which if any portions of said decoding distortion spectrum exceeds said masking threshold.

22. An apparatus according to claim 19 wherein said allocation means comprises

means for estimating a spectral envelope of said input signal;

means for comparing said spectral envelope with said rate of rolloff of frequency response;

means for identifying one or more portions of said spectral envelope which change at a rate substantially equal to or greater than said rate of rolloff of frequency response; and

means for increasing information requirements of selected encoded information corresponding to said portions of spectral envelope.

23. An apparatus according to claim 22 wherein information requirements of said selected encoded information is increased by reducing one or more portions of said masking threshold.

24. An apparatus according to claim 22 wherein information requirements of encoded information at frequencies above said one or more portions of spectral envelope are increased by an amount that decreases with increasing frequency.