Bit allocation method and device for digital audio signals using aural characteristics and signal intensities

• US 5,553,193 A
• Filed: 05/03/1993
• Issued: 09/03/1996
• Est. Priority Date: 05/07/1992
• Status: Expired due to Fees

• Current Assignee: Sony Corporation (Sony Group Corp.)
• Original Assignee: Sony Corporation (Sony Group Corp.)
• Inventors: Akagiri, Kenzo
• Primary Examiner(s): MacDonald, Allen R.
• Assistant Examiner(s): CHOWDHURY, INDRINAL
• Application Number:
  US08/056,193
• Time in Patent Office: 1,219 Days
• Field of Search: 395/2.38, 395/2.39, 395/2.43, 395/2.25, 375/1
• US Class Current: 704/200.1
• CPC Class Codes: G10L 19/0208 Subband vocoders
• CPC Class Codes: H04B 1/665 using psychoacoustic properties of the ear, e.g. masking effect

Abstract

A method for compression encoding of digital speech signals in which the digital speech signals are divided at a predetermined time interval into plural frequency bands so as to be sub-divided along both the time scale and the frequency scale to form plural sub-blocks, and data encoding is performed using an allocation bit number determined for each of said sub-blocks. The total number of bits that may be used for bit allocation is divided into at least two portions, with the bits of one of these portions being allocated in dependence upon the allowable noise level based on aural characteristics and with the bits of the remaining portion being allocated in dependence upon the intensities of signals in each sub-block. The number of bits allocated to each of the small-sized sub-blocks are determined by the sum of the bits of the two portions. In addition to the total number of bits that may be used for bit allocation, a number of bits is also set for the subsidiary information at least including the floating information and the bit length information. The aural characteristics are combined from at least the masking threshold and the minimum audibility curve and corrected by the equal-loudness curve.

Claims

1. A bit allocation method for use in digital audio signal compression in which a digital audio signal is divided into frames at a predetermined time interval and each of the frames is divided into plural frequency bands to form a set of blocks so as to divide the digital audio signal in both time and frequency, and in which compression is performed using a bit allocation number determined for each of the blocks, the method comprising steps of:

dividing a total number of bits into at least two portions for bit allocation among the blocks in the set, the portions including a first portion and a second portion, each of the portions including at least one of the total number of bits;

allocating the bits of the first portion among the blocks dependent upon an allowable noise spectrum based on aural characteristics;

calculating respective signal intensities for the blocks;

allocating the bits of the second portion among the blocks dependent upon the respective signal intensities calculated in the calculating step; and

combining the bits of the two portions together for each one of the blocks to determine the bit allocation number for the one of the blocks.

2. The method of claim 1, wherein the dividing step additionally divides the total number of bits into an additional portion, the bits divided into the additional portion being fixed in number and being for use to represent subsidiary information including block floating information and word length information.

3. The method of claim 1, wherein the step of allocating the bits of the first portion includes steps of:

calculating a masking threshold for each of the blocks, and providing a minimum audibility curve as ones of the aural characteristics;

synthesizing the masking threshold for each of the blocks and the minimum audible level curve to provide a synthesized result; and

sharing the bits of the first portion among the blocks in response to the synthesized result.

4. The method of claim 3, wherein:

the step of allocating the bits of the first portion additionally includes steps of;

providing an equal-loudness curve as an additional one of the aural characteristics, and

correcting the synthesized result according to the equal-loudness curve to provide a corrected synthesized result; and

the step of sharing the bits of the first portion among the blocks is performed in response to the corrected synthesized result.

5. The method of claim 1, wherein:

each of the blocks includes spectral components, the spectral components having respective amplitudes; and

the step of calculating the respective signal intensities for the blocks includes a step of calculating a root-mean-square value of the amplitudes of the spectral components in each of the blocks.

6. The method of claim 1, 2, or 3, wherein, in the dividing step, the total number of bits is divided into at least two portions for bit allocation among the blocks so that the first portion is larger than the second portion.

7. The method of claim 1, additionally comprising a step of dividing the digital audio signal to form the blocks by performing steps of:

dividing the digital audio signal by frequency into a low-range spectral portion, a mid-range spectral portion and a high-range spectral portion;

generating the frames by dividing the spectral portions at the predetermined time intervals of time; and

further dividing the frames by frequency into the frequency bands, the frequency bands being coincident at least in part with critical bands.

8. A method of recording a digital audio signal as a compressed signal on a recording medium, the method comprising steps of:

dividing the digital audio signal into frames at a predetermined time interval;

dividing each of the frames into plural frequency bands to form a set of blocks;

generating the compressed signal by re-quantizing the blocks using bit allocation number determined for each of the blocks, the bit allocation number being determined by steps of;

providing a total number of bits that may be used for requantizing the set of blocks,

dividing the total number of bits into at least two portions for allocation among the blocks, the portions including a first portion and a second portion, each of the portions including at least one of the total number of bits,

allocating the bits of the first portion among the blocks dependent upon an allowable noise spectrum based on aural characteristics,

calculating respective signal intensities for the blocks,

allocating the bits of the second portion dependent upon the respective signal intensities calculated in the calculating step, and

combining the bits of the two portions together for each one of the blocks to determine the bit allocation number for each one of the blocks; and

recording the compressed signal on the recording medium.

9. The recording method of claim 8, wherein, the step of recording the compressed signal on the recording medium, the compressed signal is recorded on a disc as the recording medium.

10. The recording method of claim 9, wherein, in the step of recording the compressed signal on the recording medium, the disc on which the compressed signal is recorded is an optical disc.

11. A bit allocation encoder for use in digital audio signal compression in which the digital audio signal is divided into frames at a predetermined time interval, and each of the frames is divided into plural frequency bands to form a set of blocks so as to divide the digital audio signal in both time and frequency, and in which compression is performed using a bit allocation number determined for each of the blocks, the apparatus comprising:

means for dividing a total number of bits into at least two portions for bit allocation among the blocks in the set, the portions including a first portion and a second portion, each of the portions including at least one of the total number of bits;

means for allocating the bits of the first portion among the blocks dependent upon an allowable noise spectrum based on aural characteristics;

means for calculating respective signal intensities for the blocks;

means for allocating the bits of the second portion dependent upon the respective signal intensities calculated by the calculating means; and

means for combining the bits of the two portions together for each one of the blocks to determine the bit allocation number for the one of the blocks.

12. The encoder of claim 11, wherein the aural characteristics are supplied from aural characteristics generating means at least including masking threshold generating means and minimum audibility curve generating means, the aural characteristics generating means combining an output of the masking threshold generating means and an output of the minimum audibility curve generating means to generate the aural characteristics.

13. The encoder of claim 12, wherein the aural characteristics generating means also include equal-loudness curve generating means for correcting the aural characteristics generated by combining the output of the masking threshold generating means and the output of the minimum audibility curve generating means.

14. The encoder of claim 11, additionally comprising means for encoding subsidiary information including block floating information and word length information using a set number of bits.

15. The encoder of claim 11, wherein the aural characteristics are composed of at least a masking threshold and a minimum audibility curve.

16. The encoder of claim 11, wherein the aural characteristics are composed of at least a masking threshold and a minimum audibility curve and corrected by an equal-loudness curve.

17. The encoder of claim 11, wherein the means for calculating the respective signal intensities for the blocks calculates a root-mean-square value of amplitudes in each of the blocks.

18. The encoder of claim 11, wherein the means for dividing divides the total number of bits so that the first portion is larger than the second portion.

19. The encoder of claim 11, wherein the blocks are obtained by dividing the digital audio signal into a low-range spectral portion, a mid-range spectral portion and a high-range spectral portion, by dividing the spectral portions in time into the frames, and by dividing the spectral portions in frequency into frequency bands at least in part coincident with critical bands.