# US 2010/0169089 A1 · Voice Recognizing Apparatus, Voice Recognizing Method, Voice Recognizing Program, Interference Reducing Apparatus, Interference Reducing Method, and Interference Reducing Program

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

- Number: US 2010/0169089 A1
- Document ID: US20100169089A1
- Status: Published application
- Kind: A1
- Application number: 12/087,573
- Filed: 2007-01-10
- Published: 2010-07-01
- Earliest claimed: 2006-01-11
- Assignees: NEC Corporation
- Inventors: Toru Iwasawa
- Web page: https://patentgrep.com/patent/US20100169089A1

## CPC

- G10L 21/028: using properties of sound source
- G10L 15/20: Speech recognition techniques specially adapted for robustness in adverse environments, e.g. in noise, of stress induced speech (G10L21/02 takes precedence)
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING

## Abstract

A voice recognizing apparatus includes a microphone 12 which inputs an input voice including speech voice uttered by a user speaker and interference voice uttered by an interference speaker other than the user speaker, superimposition amount determining unit 14 which determines a noise superimposition amount for the input voice on the basis of a speech voice and an interference voice separately input as the input voice, a noise superimposing unit 16 which superimposes noise according to the noise superimposition amount onto the input voice and outputs the resultant voice as noise-superimposed voice; and a voice recognizing unit 18 which recognizes the noise-superimposed voice.

## Claims

### Claim at position 1

1-48. (canceled)

### Claim 49

An interference reducing apparatus comprising:
a superimposition amount determining unit, in the case of receiving a voice including an utterance voice uttered by a user speaker and an interference signal, which estimates an interference component that interferes with recognition of said voice from information based on a position of a speaker or an utterance distance, and obtains a reduction signal of reducing influence of said interference component in said recognition of said voice by superimposing said reduction signal onto said voice; and
a reduction signal superimposing unit which superimposes said reduction signal on said voice and output the superimposed voice.

### Claim 50 (depends on 49)

The interference reducing apparatus as set forth in claim 49, further comprising a speaker position measuring unit which detects a position of a speaker whose voice should be input and outputs the position to said superimposition amount determining unit,
wherein said superimposition amount determining unit determines said reduction signal by using a distance between a user speaker obtained from said speaker position and an interference speaker as information based on the position of said speaker.

### Claim 51 (depends on 49)

The interference reducing apparatus as set forth in claim 49, further comprising an utterance distance measuring unit which measures an utterance distance between a speaker whose voice should be input and said receiving unit and outputs the obtained utterance distance to said superimposition amount determining unit,
wherein said superimposition amount determining unit determines said reduction signal by using said utterance distance as information based on the utterance distance of said speaker.

### Claim 52 (depends on 49)

The interference reducing apparatus as set forth in claim 49,
wherein said reduction signal is white noise.

### Claim 53 (depends on 49)

The interference reducing apparatus as set forth in claim 49, further comprising a voice recognizing unit which performs a voice recognizing process on said superimposed voice.

### Claim 54 (depends on 49)

A voice recognizing apparatus including the interference reducing apparatus as set forth in claim 49,
wherein an input voice has a plurality of channels, and
the voice recognizing apparatus comprises:
a plurality of superimposition amount determining units which determine a plurality of noise superimposition amount for said plurality of channels of said input voice, respectively, on the basis of said plurality of channels of said input voice;
a plurality of noise superimposing units which superimpose a plurality of noises according to said plurality of noise superimposition amounts for said plurality of channels of said input voice by said plurality of corresponding channels of said input voice, and output a plurality of noise-superimposed voices; and
a plurality of voice recognizing units which recognize said plurality of noise-superimposed voices, respectively.

### Claim 55 (depends on 54)

A voice recognizing apparatus as set forth in claim 54, further comprising a superimposition amount storing unit which previously stores an optimum noise superimposition amount associated with a relation of characteristic amounts of said utterance voice uttered by said user speaker and said interference signal,
wherein said plurality of superimposition amount determining units obtain, for each of said plurality of channels of said input voice, said respective characteristic amounts of an input voice of one of said plurality of channels as said utterance voice and input voices of the other channels as said interference signals, obtain said plurality of noise superimposition amounts corresponding to said relation of said characteristic amounts of said utterance voice and said interference signals with reference to said superimposition amount storing unit, and determine them as a plurality of noise superimposition amounts for said plurality of channels of said input voice, respectively.

### Claim 56 (depends on 53)

An interactive robot comprising the voice recognizing apparatus as set forth in claim 53,
wherein said robot executes speech and behavior according to a voice recognition result output from said voice recognizing unit.

### Claim 57 (depends on 54)

An interactive robot comprising the voice recognizing apparatus as set forth in claim 54,
wherein said robot executes speech and behavior according to a voice recognition result output from said voice recognizing unit.

### Claim 58 (depends on 54)

An interactive robot connected to a server which performs a voice recognizing process via a network,
wherein said server includes the voice recognizing apparatus as set forth in claim 54, and
said server outputs a voice recognition result output from said voice recognizing unit to said robot via said network.

### Claim 59 (depends on 54)

An interactive robot connected to a server which performs a voice recognition process via a network,
wherein said server includes the voice recognizing apparatus as set forth in claim 54, and
said server outputs a voice recognition result output from said voice recognizing unit to said robot via said network

### Claim 60

An interference reducing method comprising:
in the case of receiving a voice including an utterance voice uttered by a user speaker and an interference signal, estimating an interference component which interferes with recognition of said voice from information based on a position of a speaker or an utterance distance, and obtaining a reduction signal of reducing the influence of said interference component in said recognition of said voice by superimposing said reduction signal onto said voice; and
superimposing said reduction signal on said voice and outputting the superimposed voice.

### Claim 61 (depends on 60)

The interference reducing method as set forth in claim 60, further comprising:
detecting and outputting a position of a speaker whose voice should be input; and
determining said reduction signal by using a distance between the user speaker obtained from said speaker position and an interference speaker as information based on the position of said speaker.

### Claim 62 (depends on 60)

The interference reducing method as set forth in claim 60, further comprising:
measuring an utterance distance between the position of the speaker whose voice should be input and a voice input position and outputting the obtained utterance distance; and
determining said reduction signal by using said utterance distance as information based on the utterance distance of said speaker.

### Claim 63 (depends on 60)

The interference reducing method as set forth in claim 60, wherein said reduction signal is white noise.

### Claim 64 (depends on 60)

A voice recognizing method using the interference reducing meth as set forth in claim 60, further comprising:
performing a voice recognizing process on said superimposed voice.

### Claim 65 (depends on 60)

A voice recognizing method using the interference reducing method as set forth in claim 60,
wherein an input voice has a plurality of channels, and the voice recognizing method comprises:
determining a plurality of noise superimposition amounts for a plurality of channels of said input voice, respectively, on the basis of said plurality of channels of said input voice;
superimposing a plurality of noises according to said plurality of noise superimposition amounts for said plurality of channels of said input voice by said plurality of corresponding channels of said input voice and outputting the resultant voices as a plurality of noise-superimposed voices; and
recognizing said plurality of noise-superimposed voices, respectively.

### Claim 66 (depends on 65)

The voice recognizing method as set forth in claim 65, further comprising:
preparing a superimposition amount storing unit which previously stores an optimum noise superimposition amount associated with the relation of characteristic amounts of said utterance voice uttered by said user speaker and said interference signals;
obtaining, for each said plurality of channels of said input voice, said respective characteristic amounts of one of said plurality of channels of said input voice as said utterance voice and input voices of the other channels as said interference signals by said plurality of channels of said input voice;
obtaining said plurality of noise superimposition amounts corresponding to said relation of said characteristic amounts of said utterance voice and said interference signals with reference to said superimposition amount storing unit; and
determining them as a plurality of noise superimposition amounts for said plurality of channels of said input voice, respectively.

### Claim 67

An interference reducing program which makes a computer function as:
in the case of receiving a voice including an utterance voice uttered by a user speaker and an interference signal,
a unit which estimates an interference component which interferes with recognition of said voice from information based on a position of a speaker or an utterance distance, and obtains a reduction signal of reducing the influence of said interference component in said recognition of said voice by superimposing said reduction signal onto said voice; and
a unit which superimposes said reduction signal on said voice and outputs the superimposed voice.

### Claim 68 (depends on 67)

The interference reducing program as set forth in claim 67,
wherein said computer is further made function as a unit which detects a position of a speaker whose voice should be input and outputs the position, and
said unit which obtains said reduction signal determines said reduction signal by using a distance between a user speaker obtained from said speaker position and an interference speaker as information based on the position of said speaker.

### Claim 69 (depends on 67)

The interference reducing program as set forth in claim 67,
wherein said computer is further made to function as a unit which measures an utterance distance between a position of a speaker whose voice should be input and a voice input position and outputs the obtained utterance distance, and
a unit which obtains said reduction signal determines said reduction signal determines said reduction signal by using utterance distance as information based on the utterance distance of said speaker.

### Claim 70 (depends on 67)

The interference reducing program as set forth in claim 67,
wherein said reduction signal is white noise.

### Claim 71 (depends on 67)

A voice recognizing program including the interference reducing program as set forth in claim 67, which makes a computer function as a unit which performs a voice recognizing process on said superimposed voice.

### Claim 72 (depends on 67)

A voice recognizing program including the interference reducing program as set forth in claim 67,
wherein an input voice has a plurality of channels, and
said program makes said computer function as:
a unit which determines a plurality of noise superimposition amounts for said plurality of channels of said input voice, respectively, on the basis of said plurality of channels of said input voice;
a unit which superimposes a plurality of noises according to said plurality of noise superimposition amounts for said plurality of channels of said input voice by said plurality of corresponding channels of said input voice by said plurality of corresponding channels of said input voice and outputs a plurality of noise-superimposed voices; and
a unit which recognizes each of said plurality of noise-superimposed voices, respectively.

### Claim 73 (depends on 72)

The voice recognizing program as set forth in claim 72,
wherein said computer comprises a superimposition amount storing unit which previously stores an optimum noise superimposition amount associated with the relation of characteristic amounts of said utterance voice uttered by said user speaker and said interference signal, and
said noise superimposition amount determining unit obtains, for each of said plurality of channels of said input voice, said respectively characteristic amounts of an input voice of one of said plurality of channels of said input voice as said utterance voice and input voices of the other channels as said interference signals, obtains said plurality of noise superimposition amounts corresponding to said relation of said characteristic amounts of said utterance voice and said interference signals with reference to said superimposition amount storing unit, and determines them as a plurality of noise superimposition amounts for said plurality of channels of said input voice, respectively.
