Quantization bit number allocation by first selecting a subband signal having a maximum of signal to mask ratios in an input signal

• US 5,469,474 A
• Filed: 06/24/1993
• Issued: 11/21/1995
• Est. Priority Date: 06/24/1992
• Status: Expired due to Term

• Current Assignee: NEC Electronics Corporation (Renesas Electronics Corporation)
• Original Assignee: NEC Corporation
• Inventors: Kitabatake, Osamu
• Primary Examiner(s): Chin, Stephen
• Assistant Examiner(s): PHAN, HAI
• Application Number:
  US08/080,766
• Time in Patent Office: 880 Days
• Field of Search: 375/243, 375/254, 375/240, 375/242, 375/244, 395/2.38, 381/29, 381/36, 381/31, 341/143
• US Class Current: 375/243
• CPC Class Codes: H04B 1/665 using psychoacoustic properties of the ear, e.g. masking effect

Abstract

As a selected subband signal, selected is one of frequency band signals of an input signal that has a maximal signal to mask ratio. The selected subband signal is assigned, as a primary provisional bit number, with a maximal quantization bit number allocatable thereto. A primary mask to noise ratio of the selected subband signal is calculated. Other subband signals are assigned with temporary bit numbers incremented one by one from zero with secondary mask to noise ratios of the other subband signal meanwhile calculated each time when the temporary bit numbers are incremented by one. The temporary bit numbers are allocated to the other subband signals as secondary provisional bit numbers when the secondary mask to noise ratios individually exceed the primary mask to noise ratio. Some of the secondary provisional bit numbers may be equal to zero. A sum of the primary and the secondary provisional bit numbers is compared with a total of quantization bit numbers allocatable to the frequency band signals. If the sum does not exceed the total, the primary and the secondary provisional bit numbers are allocated to the selected and the other subband signals. If the sum exceeds the total, the primary and the secondary provisional bit numbers are uniformly reduced and allocated, unless reduced to negative, to the selected and the other subband signals. Zero is used instead of at least one of the secondary provisional bit numbers that is reduced to negative.

Claims

1. A quantization bit number allocating method for allocating optimum quantization bit numbers to a plurality of frequency subband signals, said method comprising the steps of:

selecting, as a selected subband signal, one of said plurality of frequency subband signals having a maximum ratio of a maximal signal level to a mask level;

allocating to said selected subband signal, as a primary provisional bit number, a maximal quantization bit number allocatable to said selected subband signal;

calculating a primary mask-to-noise ratio of said selected subband signal;

setting a provisional bit number to each remaining frequency subband signal of said plurality of frequency subband signals which has not been selected;

incrementing, one by one starting from zero, each of the provisional bit numbers set to said remaining frequency subband signals of said plurality of frequency subband signals, meanwhile calculating secondary mask-to-noise ratios for said remaining frequency subband signals of said plurality of frequency subband signals each time an associated provisional bit number of said provisional bit numbers is incremented by one;

allocating said provisional bit numbers to said frequency subband signals which have not been selected as secondary provisional bit numbers when said secondary mask-to-noise ratios individually exceed said primary mask-to-noise ratio;

summing up said primary and said secondary provisional bit numbers into a sum;

comparing said sum with a total of quantization bit numbers allocatable to said frequency subband signals;

using said primary and said secondary provisional bit numbers collectively as said optimal quantization bit numbers if said sum does not exceed said total; and

if said sum does exceed said total, performing the following steps of;

calculating a quotient by dividing a numerator difference, said numerator difference equals to said sum minus said total, by a denominator difference, said denominator difference equals to a predetermined integer minus the number of said remaining frequency subband signals, excluding at least one particular subband signal from said remaining frequency subband signals;

calculating bit number differences by subtracting an integral part of said quotient from said primary and secondary provisional bit numbers; and

using said bit number differences as said optimal quantization bit numbers if said bit number differences are not negative.

2. A quantization bit number allocating method as claimed in claim 1, wherein zero is used as one of said optimal quantization bit numbers and is allocated to at least one particular frequency subband signal of said plurality of frequency subband signals if the secondary mask-to-noise ratio associated with said at least one particular frequency subband signal exceeds said primary mask-to-noise ratio before the corresponding provisional bit number is incremented from zero to one.

3. A quantization bit number allocating method as claimed in claim 1, wherein zero is used as at least one of said optimal quantization bit numbers instead of one of said bit number differences that is negative.

4. A quantization bit number allocating device for allocating optimal quantization bit numbers to a plurality of frequency subband signals, said device comprising:

first means for selecting, as a selected subband signal, one of said frequency subband signals that has a maximum ratio of a maximal signal level to a mask level;

second means for allocating to said selected subband signal, as a primary provisional bit number, a maximal quantization bit number allocatable to said selected subband signal;

third means for calculating a primary mask-to-noise ratio of said selected subband signal;

a fourth means for setting a provisional bit number to each remaining frequency subband signal of said plurality of frequency subband signals which has not been selected;

fifth means for incrementing, one by one starting from zero, each of the provisional bit numbers set to the frequency subband signals of said plurality of frequency subband signals which has not been selected, and for meanwhile calculating secondary mask-to-noise ratios for the remaining frequency subband signals of said plurality of frequency subband signals each time an associated provisional bit number of said provisional bit numbers is incremented by one;

sixth means for allocating said provisional bit numbers to said frequency subband signals which have not been selected as secondary provisional bit numbers when said secondary mask-to-noise ratios individually exceed said primary mask-to-noise ratio;

seventh means for summing up said primary and said secondary provisional bit numbers into a sum;

eighth means for comparing said sum with a total of quantization bit numbers allocatable to said frequency subband signals; and

ninth means for allocating said primary and said secondary provisional bit numbers collectively as said optimal quantization bit numbers to said selected and said other frequency subband signals if said sum does not exceed said total;

tenth means for calculating a quotient by dividing a numerator difference which equals said sum minus said total by a denominator difference which equals said plurality minus the number of other frequency subband signals, said tenth means only being used when said sum exceeds said total;

eleventh means for calculating bit number differences by subtracting an integer part of said quotient from said primary and secondary provisional bit numbers, said eleventh means only being used when said sum exceeds said total; and

twelfth means for allocating said bit number differences as said optimal quantization bit numbers if said bit number differences are not negative, said twelfth means only being used when said sum exceeds said total.

5. A quantization bit number allocating device as claimed in claim 4, wherein said sixth means allocates zero to at least one particular frequency subband signal of said plurality of frequency subband signals if the secondary mask-to-noise ratio associated with said at least one particular frequency subband signal exceeds said primary mask-to-noise ratio before the corresponding provisional bit number is incremented from zero to one.