# US 2001/0027391 A1 · Excitation vector generator, speech coder and speech decoder

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2001/0027391 A1
- Document ID: US20010027391A1
- Status: Published application
- Kind: A1
- Application number: 09/843,877
- Filed: 2001-04-30
- Published: 2001-10-04
- Earliest claimed: 1996-11-07
- Assignees: Matsushita Electric Industrial Co., Ltd.
- Inventors: Kazutoshi Yasunaga; Toshiyuki Morii; Hiroyuki Ehara
- Web page: https://patentgrep.com/patent/US20010027391A1

## CPC

- G10L 19/135: Vector sum excited linear prediction [VSELP]
- G10L 19/12: the excitation function being a code excitation, e.g. in code excited linear prediction [CELP] vocoders
- G10L 2019/0007: {Codebook element generation}
- G10L 2019/0013: {Codebook search algorithms}
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING

## Abstract

A random code vector reading section and a random codebook of a conventional CELP type speech coder/decoder are respectively replaced with an oscillator for outputting different vector streams in accordance with values of input seeds, and a seed storage section for storing a plurality of seeds . This makes it unnecessary to store fixed vectors as they are in a fixed codebook (ROM), thereby considerably reducing the memory capacity.

## Claims

### Claim 1

An excitation vector generator, comprising:
a providing system that provides an input vector having at least one pulse, each pulse of said at least one pulse having a predetermined position and a predetermined polarity;
a storage system that stores at least one fixed waveform; and
a convolution system that enables modification of said input vector with said at least one fixed waveform to transform a waveform of said input vector, said convoluting system outputting said transformed input vector as an excitation vector to improve a speech quality when a target for a random codebook search is encoded with said input vector.

### Claim 2 (depends on 1)

The excitation vector generator of claim 1, wherein said input vector comprises a sparse vector.

### Claim 3 (depends on 1)

The excitation vector generator of claim 1, wherein said input vector is provided from an algebraic codebook.

### Claim 4 (depends on 1)

The excitation vector generator of claim 1, wherein said input vector comprises a vector having a plurality of non-zero samples.

### Claim 5 (depends on 1)

The excitation vector generator of claim 1, wherein said convolution system performs a convolution using one fixed waveform of said at least fixed waveform that is read from said storage system.

### Claim 6 (depends on 1)

The excitation vector generator of claim 1, wherein said convolution system spreads an energy distribution of said input vector over a subframe.

### Claim 7 (depends on 1)

The excitation vector generator of claim 1, wherein said at least one fixed waveform comprises three different fixed waveforms.

### Claim 8 (depends on 1)

The excitation vector generator of claim 1, wherein said at least one fixed waveform comprises three different fixed waveforms having a different amount of energy spreading.

### Claim 9

An excitation vector generator, comprising:
a providing system that provides an input vector having a plurality of non-zero samples;
a storage system that stores at least one fixed waveform; and
a convolution system that transforms said input vector with said at least one fixed waveform to enable a modification of an energy distribution of said input vector, said convolution system outputting said transformed input vector as an excitation vector to improve a speech quality when a target for a random codebook search is encoded with the input vector.

### Claim 10 (depends on 9)

The excitation vector generator of claim 9, wherein said convolution system disperses said energy distribution of said input vector.

### Claim 11 (depends on 9)

The excitation vector generator of claim 9, wherein said energy distribution is modified by spreading an energy of each non-zero sample of said plurality of non-zero samples over each sample adjacent to said plurality of non-zero samples.

### Claim 12 (depends on 9)

The excitation vector generator of claim 9, wherein said energy distribution is modified by spreading an energy of each non-zero sample of said plurality of non-zero samples around each of said plurality of non-zero samples.

### Claim 13 (depends on 9)

The excitation vector generator of claim 9, wherein said energy distribution is modified by spreading an energy of each non-zero sample of said plurality of non-zero samples over each area adjacent to said plurality of non-zero samples.

### Claim 14 (depends on 9)

The excitation vector generator of claim 9, wherein said convolution system performs a convolution using a fixed waveform read from said storage system.

### Claim 15 (depends on 9)

The excitation vector generator of claim 9, wherein said convolution system spreads an energy distribution of said input vector over a subframe.

### Claim 16 (depends on 9)

The excitation vector generator of claim 9, wherein said at least one fixed waveform comprises three fixed waveforms, each fixed waveform of said three fixed waveforms having a different waveform.

### Claim 17 (depends on 9)

The excitation vector generator of claim 9, wherein said at least one fixed waveform comprises thee fixed waveforms, each fixed waveform of said three fixed waveforms having a different amount of energy spreading from one another.

### Claim 18

A method of generating an excitation vector, comprising:
receiving a code number corresponding to at least one position;
providing an input vector corresponding to the received code number;
reading out at least one pre-stored fixed waveform from a storage system;
convolution processing the input vector and the at least one fixed waveform to generate an excitation vector; and
outputting the generated excitation vector to improve a speech quality when a target for a random codebook search is encoded with the input vector.

### Claim 19 (depends on 18)

The method of claim 18, wherein providing an input vector comprises providing a sparse vector.

### Claim 20

A method for generating an excitation vector, comprising:
providing an input vector having at least one pulse, each pulse of the at least one pulse having a predetermined position and a predetermined polarity;
storing at least one fixed waveform; and
convoluting the input vector with the at least one fixed waveform so that a transformed excitation vector is produced, the transformed excitation vector being output to improve a speech quality when a target for a random codebook search is encoded with the input vector.

### Claim 21

A method for generating an excitation vector, comprising:
providing an input vector having a plurality of non-zero samples;
storing at least one fixed waveform; and
convoluting the input vector with the at least one fixed waveform to enable a modification of an energy distribution of the input vector, which is output as an excitation vector to improve a speech quality when a target for a random codebook search is encoded with the input vector.
