Encoding device and encoding method
First Claim
Patent Images
1. A speech coding method comprising:
- performing a modified discrete cosine transformation for an input speech signal to provide spectrum data;
performing algebraic vector quantization coding using the spectrum data at a bit rate of 4 kbit/s or 12 kbit/s, the algebraic vector quantization coding comprising;
calculating a global gain for the spectrum data;
dividing the spectrum data into a plurality of subbands;
correcting the spectrum data of each subband of the plurality of subbands;
transforming the corrected spectrum data into a lattice vector;
calculating index information from the lattice vector; and
multiplexing the global am with the index information to generate coded information, and outputting the coded information, which is related to the speech signal to be transmitted to a speech decoding apparatus,wherein the correcting calculates an average value of an amplitude of spectrum data for each subband, and assigns zero to a value of a sample having an amplitude equal to or less than the average value, among the group of samples related to the spectrum data of each subband,wherein at least one of the performing modified discrete cosine transformation, performing algebraic vector quantization coding, calculating a global gain, dividing, correcting, transforming, calculating index information and multiplexing is performed by a processor.
3 Assignments
0 Petitions
Accused Products
Abstract
An encoding device and encoding method improve a quality of a decoded signal under very low bit rate conditions using a small amount of computation. A spectrum corrector performs correction processing on a subspectrum in each subband in such a manner that samples equal to or greater than a subspectrum average value are left unchanged and samples smaller than the subspectrum average value are replaced by zero. As a result of this, it is possible to significantly reduce the number of bits required to quantize the subspectrums without a substantial reduction in quality in a local searcher and in a multi-rate indexer.
-
Citations
8 Claims
-
1. A speech coding method comprising:
-
performing a modified discrete cosine transformation for an input speech signal to provide spectrum data; performing algebraic vector quantization coding using the spectrum data at a bit rate of 4 kbit/s or 12 kbit/s, the algebraic vector quantization coding comprising; calculating a global gain for the spectrum data; dividing the spectrum data into a plurality of subbands; correcting the spectrum data of each subband of the plurality of subbands; transforming the corrected spectrum data into a lattice vector; calculating index information from the lattice vector; and multiplexing the global am with the index information to generate coded information, and outputting the coded information, which is related to the speech signal to be transmitted to a speech decoding apparatus, wherein the correcting calculates an average value of an amplitude of spectrum data for each subband, and assigns zero to a value of a sample having an amplitude equal to or less than the average value, among the group of samples related to the spectrum data of each subband, wherein at least one of the performing modified discrete cosine transformation, performing algebraic vector quantization coding, calculating a global gain, dividing, correcting, transforming, calculating index information and multiplexing is performed by a processor.
-
-
2. A speech coding apparatus comprising:
-
a processor; and a memory storing instructions, wherein the processor performs the instructions stored in the memory, and comprises; an orthogonal transformer that performs a modified discrete cosine transformation for an input speech signal to provide spectrum data; and an algebraic vector quantization encoder that performs algebraic vector quantization coding using the spectrum data at a bit rate of 4 kbit/s or 12 kbit/s, the algebraic vector quantization encoder comprising; a global gain calculator that calculates the spectrum data, a spectrum corrector that divides the spectrum data into a plurality of subbands and corrects the spectrum data of each subband of the plurality of subbands, a transformer that transforms the corrected spectrum data into a lattice vector, a multi-rate indexer that calculates index information from the lattice vector, and a multiplexer that multiplexes the global gain with the index information to Generate coded information, and outputs the coded information, which is related to the speech signal to be transmitted to a speech decoding apparatus, wherein the spectrum corrector evaluates a magnitude of an amplitude of spectrum data for each subband, selects a predetermined number of samples in a descending order of the magnitude of the amplitude, among a group of samples related to the spectrum data of each subband, and assigns zero to the value of the sample other than the selected predetermined number of samples.
-
-
3. A speech coding apparatus comprising:
-
a processor; and a memory storing instructions, wherein the processor performs the instructions stored in the memory, and comprises; an orthogonal transformer that performs a modified discrete cosine transformation for an input speech signal to provide spectrum data; and an algebraic vector quantization encoder that performs algebraic vector quantization coding using the spectrum data at a bit rate of 4 kbit/s or 12 kbit/s, the algebraic vector quantization encoder comprising; a global rum data, a spectrum corrector that divides the spectrum data into a plurality of subbands and corrects the spectrum data of each subband of the plurality of subbands, a transformer that transforms the corrected spectrum data into a lattice vector, a multi-rate indexer that calculates index information from the lattice vector, and a multiplexer that multiplexes the global gain with the index information to generate coded information and out uts the coded information which is related to the speech signal to be transmitted to a speech decoding apparatus, wherein the spectrum corrector calculates an average value of an amplitude of spectrum data for each subband and assigns zero to a value of a sample having an amplitude equal to or less than the average value, among a group of samples related to the spectrum data of each subband. - View Dependent Claims (4, 5, 6, 7, 8)
-
Specification