Coding apparatus, decoding apparatus, coding method and decoding method
First Claim
1. A coding apparatus comprising:
- an acquisition section that acquires a spectrum of a voice signal including a low frequency band spectrum and a high frequency band spectrum;
a first coding section that codes the low frequency band spectrum, to generate first coded information;
a first decoding section that decodes the first coded information, to generate decoded low frequency band spectrum;
a generation section that generates a shape of an estimated high frequency band spectrum by duplicating the decoded low frequency band spectrum;
a first subband amplitude calculation section that divides the estimated high frequency band spectrum into a plurality of subbands and calculates a first average amplitude value of the estimated high frequency band spectrum in each of the plurality of subbands;
a second subband amplitude calculation section that divides the high frequency band spectrum into a plurality of subbands and calculates a second average amplitude value of the high frequency band spectrum in each of the plurality of subbands;
a gain codebook that outputs only a gain of specific subbands, the specific subbands including a subband of the estimated high frequency band spectrum adjacent to the low frequency band spectrum among the plurality of subbands of the estimated high frequency band spectrum;
an interpolation section that finds a gain of a subband of the estimated high frequency band spectrum other than the specific subbands by interpolating the gain of the specific subbands;
a multiplication section that multiplies the first average amplitude value by the gain output from the gain codebook or the interpolation section in each of the subbands;
a second coding section that calculates a distortion between a multiplied result of the multiplication section and the second average amplitude value and codes only the gain of the specific subbands when the distortion is smallest; and
an output terminal that outputs coded information obtained by the first and second coding sections.
2 Assignments
0 Petitions
Accused Products
Abstract
There is disclosed an encoding device capable of improving similarity between the high frequency band spectrum of the original signal and a new spectrum to be generated while realizing a low bit rate when encoding a wide-band signal spectrum. The encoding device has sub-band amplitude calculation units for calculating the amplitude of the respective sub-bands for the high frequency band spectrum obtained from the wide-band signal. A search unit and a gain codebook select some sub-bands from a plurality of sub-bands and only the gain of the selected sub-bands is subjected to encoding. An interpolation unit expresses the gain of the sub-band not selected, by mutually interpolating the selected gains.
-
Citations
11 Claims
-
1. A coding apparatus comprising:
-
an acquisition section that acquires a spectrum of a voice signal including a low frequency band spectrum and a high frequency band spectrum; a first coding section that codes the low frequency band spectrum, to generate first coded information; a first decoding section that decodes the first coded information, to generate decoded low frequency band spectrum; a generation section that generates a shape of an estimated high frequency band spectrum by duplicating the decoded low frequency band spectrum; a first subband amplitude calculation section that divides the estimated high frequency band spectrum into a plurality of subbands and calculates a first average amplitude value of the estimated high frequency band spectrum in each of the plurality of subbands; a second subband amplitude calculation section that divides the high frequency band spectrum into a plurality of subbands and calculates a second average amplitude value of the high frequency band spectrum in each of the plurality of subbands; a gain codebook that outputs only a gain of specific subbands, the specific subbands including a subband of the estimated high frequency band spectrum adjacent to the low frequency band spectrum among the plurality of subbands of the estimated high frequency band spectrum; an interpolation section that finds a gain of a subband of the estimated high frequency band spectrum other than the specific subbands by interpolating the gain of the specific subbands; a multiplication section that multiplies the first average amplitude value by the gain output from the gain codebook or the interpolation section in each of the subbands; a second coding section that calculates a distortion between a multiplied result of the multiplication section and the second average amplitude value and codes only the gain of the specific subbands when the distortion is smallest; and an output terminal that outputs coded information obtained by the first and second coding sections. - View Dependent Claims (2, 3, 4, 6, 8)
-
-
5. A decoding apparatus that decodes coded information relating to a spectrum of a voice signal including a low frequency band spectrum and a high frequency band spectrum, comprising:
-
a first decoding section that decodes coded information relating to the low frequency band spectrum, to generate a decoded low frequency band spectrum; a second decoding section that decodes coded information relating to the high frequency band spectrum, to generate only a gain of specific subbands, the specific subbands including a subband of the high frequency band spectrum adjacent to the low frequency band spectrum among the plurality of subbands into which the high frequency band spectrum is divided; an output terminal that outputs the decoded low frequency band spectrum obtained by the first decoding section and the gains obtained by the second decoding section; an interpolation section that finds a gain of a subband other than the specific subbands by interpolating the gain of the specific subbands; and a high frequency band spectrum generation section that generates the high frequency band spectrum using the decoded low frequency band spectrum obtained by the first decoding section and the gains obtained by the second decoding section and the interpolation section. - View Dependent Claims (7, 9)
-
-
10. A coding method comprising:
-
an acquisition step of acquiring a spectrum of a voice signal including a low frequency band spectrum and a high frequency band spectrum; a first coding step of coding the low frequency band spectrum, to generate first coded information; a first decoding step of decoding the first coded information, to generate decoded low frequency hand spectrum; a generation step of generating a shape of an estimated high frequency band spectrum by duplicating the decoded low frequency band spectrum; a first subband amplitude calculation step of dividing the estimated high frequency band spectrum into a plurality of subbands and calculating a first average amplitude value of the estimated high frequency band spectrum in each of the plurality of subbands; a second subband amplitude calculation step of dividing the high frequency band spectrum into a plurality of subbands and calculating a second average amplitude value of the high frequency band spectrum in each of the plurality of subbands; a gain output step of outputting only a gain of specific subbands, the specific subbands including a subband of the estimated high frequency band spectrum adjacent to the low frequency band spectrum among the plurality of subbands of the estimated high frequency hand spectrum; an interpolation step of finding a gain of a subband of the estimated high frequency band spectrum other than the specific subbands by interpolating the gain of the specific subbands; a multiplication step of multiplying the first average amplitude value by the gain output from the gain output step or the interpolation step in each of the subbands; a second coding step of calculating a distortion between a multiplied result of the multiplication step and the second average amplitude value and coding only the gain of the specific subbands when the distortion is smallest; and an output step of outputting coded information obtained by the first and second coding steps.
-
-
11. A decoding method that decodes coded information relating to a spectrum of a voice signal including a low frequency band spectrum and a high frequency band spectrum, comprising:
-
a first decoding step of decoding coded information relating to the low frequency band spectrum, to generate a decoded low frequency band spectrum; a second decoding step of decoding coded information relating to the high frequency band spectrum, to generate only a gain of specific subbands, the specific subbands including a subband of the high frequency band spectrum adjacent to the low frequency band spectrum among the plurality of subbands into which the high frequency band spectrum is divided; an interpolation step of finding a gain of a subband other than the specific subbands by interpolating the gain of the specific subbands; and a high frequency band spectrum generation step of generating the high frequency band spectrum using the decoded low frequency band spectrum obtained by the first decoding step and the gains obtained by the second decoding step and the interpolation step.
-
Specification