Apparatus and method for correcting the difference in frequency characteristics between microphones for analyzing speech and for creating a recognition dictionary

• US 6,032,115 A
• Filed: 09/26/1997
• Issued: 02/29/2000
• Est. Priority Date: 09/30/1996
• Status: Expired due to Term

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: Kanazawa, Hiroshi, Isaka, Takehiko, Tsuboi, Hiroyuki, Nagata, Yoshifumi
• Primary Examiner(s): Hudspeth, David R.
• Assistant Examiner(s): Storm, Donald L.
• Application Number:
  US08/935,082
• Time in Patent Office: 886 Days
• Field of Search: 704/244, 704/248, 704/233, 704/253, 704/224, 704/231, 704/234, 704/243, 704/245, 704/200
• US Class Current: 704/234
• CPC Class Codes: G10L 15/065 Adaptation
• CPC Class Codes: 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

Abstract

In sound recognition apparatus of the present invention, user'"'"'s utterance or a sound provided by an output section using previously stored sound waveforms is simultaneously inputted through a basic microphone of known frequency characteristics and an input microphone of unknown frequency characteristics. An analysis section respectively analyzes the frequency of the input speech through the basic microphone and the input microphone. A frequency characteristics calculation section calculates first difference data between the frequencies of the input speech of the basic microphone and the input microphone, and calculates frequency characteristics of the input microphone according to the first difference data and the frequency characteristics of the basic microphone. A frequency characteristics correction section calculates second difference data between the frequency characteristics of the input microphone and known frequency characteristics of a dictionary data microphone, and corrects input speech to be recognized through the input microphone as speech data of the frequency characteristics of the dictionary data microphone according to the second difference data. A recognition section recognizes corrected speech data by referring to a recognition dictionary storing data previously created through the dictionary data microphone.

Claims

1. A sound recognition apparatus, comprising:

an input-means for simultaneously inputting speech through a basic microphone of known frequency characteristics and an input microphone of unknown frequency characteristics, the basic microphone being different from a dictionary data microphone of known frequency characteristics which collected data for previously creating a recognition dictionary;

an analysis means for respectively analyzing the frequency of the input speech through the basic microphone and the input microphone;

a frequency characteristics measurement means for calculating as a first difference data, the difference between the frequencies of the input speech of the basic microphone and the input microphone, and for calculating frequency characteristics of the input microphone according to the first difference data and the frequency characteristics of the basic microphone;

a frequency characteristics correction means for calculating as a second difference data, the difference between the frequency characteristics of the input microphone and the known frequency characteristics of the dictionary data microphone which collects data for creating a recognition dictionary, and for correcting input speech to be recognized through the input microphone as speech data of the frequency characteristics of the dictionary data microphone according to the second difference data; and

a recognition means for recognizing corrected speech data by referring to the recognition dictionary storing data previously created through the dictionary data microphone.

2. The sound recognition apparatus according to claim 1,

wherein said input means respectively extracts speech data between a starting point and an ending point from the input speech through the basic microphone and the input microphone.

3. The sound recognition apparatus according to claim 2,

wherein said analysis means calculates average spectral vectors from extracted speech data of the basic microphone and the input microphone.

4. The sound recognition apparatus according to claim 3,

wherein said frequency characteristics measurement means calculates the first difference data between the average spectral vector corresponding to input speech of the basic microphone and the average spectral vector corresponding to input speech of the input microphone.

5. The sound recognition apparatus according to claim 1,

wherein said frequency characteristics correction means corrects frequency parameter of the input speech to be recognized by unit of predetermined timing as a subtraction between the frequency parameter and the second difference data.

6. The sound recognition apparatus according to claim 5,

wherein said frequency characteristics correction means converts the corrected frequency parameter to a speech feature vector to be matched with the recognition dictionary.

7. A sound recognition apparatus, comprising:

a sound output means for outputting sound using previously stored sound waveforms;

an input means for simultaneously inputting sound from said sound output means through a basic microphone of known frequency characteristics and an input microphone of unknown frequency characteristics, the basic microphone being different from a dictionary data microphone of known frequency characteristics which collected data for previously creating a recognition dictionary;

an analysis means for respectively analyzing the frequencies of the input sound through the basic microphone and the input microphone;

a frequency characteristics measurement means for calculating as a first difference data, the difference between the frequencies of the input sound of the basic microphone and the input microphone, and for calculating frequency characteristics of the input microphone according to the first difference data and the frequency characteristics of the basic microphone;

a frequency characteristics correction means for calculating as a second difference data, the difference between the frequency characteristics of the input microphone and the known frequency characteristics of the dictionary data microphone, and for correcting input sound to be recognized through the input microphone as sound data of the frequency characteristics of the dictionary data microphone according to the second difference data; and

a recognition means for recognizing corrected sound data by referring to the recognition dictionary storing data previously created through the dictionary data microphone.

8. A method for recognizing sound, comprising the steps of:

inputting speech through a basic microphone of known frequency characteristics and an input microphone of unknown frequency characteristics simultaneously, the basic microphone being different from a dictionary data microphone of known frequency characteristics which collected data for previously creating a recognition dictionary;

analyzing the frequencies of the input speech through the basic microphone and the input microphone respectively;

calculating as a first difference data, the difference between the frequencies of the input speech of the basic microphone and the input microphone;

calculating frequency characteristics of the input microphone according to the first difference data and the known frequency characteristics of the basic microphone;

calculating as a second difference data, the difference data between the frequency characteristics of the input microphone and the known frequency characteristics of the dictionary data microphone;

correcting input speech to be recognized through the input microphone as speech data of the frequency characteristics of the dictionary data microphone according to the second difference data; and

recognizing corrected speech data by referring to the recognition dictionary storing data previously created through the dictionary data microphone.

9. A method for recognizing sound, comprising steps of:

outputting sound using previously stored sound waveforms;

inputting the sound through a basic microphone of known frequency characteristics and an input microphone of unknown frequency characteristics simultaneously, the basic microphone being different from a dictionary data microphone of known frequency characteristics which collected data for previously creating a recognition dictionary;

analyzing frequencies of the input sound through the basic microphone and the input microphone respectively;

calculating as a first difference data, the difference between the frequencies of the input sound of the basic microphone and the input microphone;

calculating frequency characteristics of the input microphone according to the first difference data and the known frequency characteristics of the basic microphone;

calculating as a second difference data, the difference between the frequency characteristics of the input microphone and the known frequency characteristics of the dictionary data microphone;

correcting input sound to be recognized through the input microphone as sound data of the frequency characteristics of the dictionary data microphone according to the second difference data; and

recognizing corrected sound data by referring to the recognition dictionary storing data previously created through the dictionary data microphone.

10. A computer readable memory containing computer readable instructions, comprising:

an instruction means for causing a computer to receive speech through a basic microphone of known frequency characteristics and an input microphone of unknown frequency characteristics simultaneously, the basic microphone being different from a dictionary data microphone of known frequency characteristics which collected data for previously creating a recognition dictionary;

an instruction means for causing a computer to respectively analyze the frequency of the input speech through the basic microphone and the input microphone;

an instruction means for causing a computer to calculate as a first difference data, the difference between the frequencies of the input speech of the basic microphone and the input microphone;

an instruction means for causing a computer to calculate frequency characteristics of the input microphone according to the first difference data and the known frequency characteristics of the basic microphone;

an instruction means for causing a computer to calculate as a second difference data, the difference between the frequency characteristics of the input microphone and the known frequency characteristics of the dictionary data microphone;

an instruction means for causing a computer to correct input speech to be recognized through the input microphone as speech data of the frequency characteristics of the dictionary data microphone according to the second difference data; and

an instruction means for causing a computer to recognize corrected speech data by referring to the recognition dictionary storing data previously created through the dictionary data microphone.

11. A computer readable memory containing computer readable instructions, comprising:

an instruction means for causing a computer to output sound using previously stored sound waveform;

an instruction means for causing a computer to receive the sound through a basic microphone of known frequency characteristics and an input microphone of unknown frequency characteristics simultaneously, the basic microphone being different from a dictionary data microphone of known frequency characteristics which collected data for previously creating a recognition dictionary;

an instruction means for causing a computer to respectively analyze the frequency of the input sound through the basic microphone and the input microphone;

an instruction means for causing a computer to calculate as a first difference data, the difference between the frequencies of the input sound of the basic microphone and the input microphone;

an instruction means for causing a computer to calculate frequency characteristics of the input microphone according to the first difference data and the frequency characteristics of the basic microphone;

an instruction means for causing a computer to calculate as a second difference data, the difference between the frequency characteristics of the input microphone and the known frequency characteristics of the dictionary data microphone;

an instruction means for causing a computer to correct input sound to be recognized through the input microphone as sound data of the frequency characteristics of the dictionary data microphone according to the second difference data; and

an instruction means for causing a computer to recognize corrected sound data by referring to the recognition dictionary storing data previously created through the dictionary data microphone.