Content analyzer mixing inverse-direction-probability-weighted noise to input signal

• US 5,828,997 A
• Filed: 06/07/1995
• Issued: 10/27/1998
• Est. Priority Date: 06/07/1995
• Status: Expired due to Fees

• Current Assignee: Sensimetrics Corporation
• Original Assignee: Sensimetrics Corporation
• Inventors: Durlach, Nathaniel I., Berkovitz, Robert A.
• Primary Examiner(s): Hudspeth, David R.
• Assistant Examiner(s): Smits, Talivaldis Ivars
• Application Number:
  US08/485,368
• Time in Patent Office: 1,238 Days
• Field of Search: 367/124, 367/125, 367/126, 367/129, 342/92, 704/233
• US Class Current: 704/233
• CPC Class Codes: G01H 3/08 Analysing frequencies present in complex vibrations, e.g. comparing harmonics present acoustic presence detection G01V1/001
• CPC Class Codes: G01V 1/001 Acoustic presence detection
• CPC Class Codes: G10L 2021/02166 Microphone arrays; Beamforming
• CPC Class Codes: G10L 21/0208 Noise filtering

Abstract

A speech recognition system combines content analysis of a signal with directional analysis of the signal to calculate probable contents of the signal and the probability that the contents are as calculated. Content analysis can be combined with independent source direction analysis, consisting of the probable direction of the source and the probability that the source is as determined, to refine the content analysis. The direction analysis can affect either the calculation of the probable contents of the signal or the calculation of the probability that the contents are as calculated. The improved information obtained by combining the content analysis with directional analysis can be used to improve subsequent content analyses or directional analyses.

Claims

1. A recognition system for a source signal comprising:

input means for obtaining at least one input signal for the source signal;

a direction analyzer having a direction input coupled to the input means, a direction output, and a direction processor for calculating a probability that the source signal originates from one of at least one predetermined direction and providing to the direction output a direction probability signal representative of the probability;

a noise generator having an input coupled to the direction output and a noise output, for providing a noise signal inversely proportional to the probability;

a mixer having a first mixer input coupled to the input means, a second mixer input coupled to the noise output, and a mixer output, for combining signals on the first mixer input and the second mixer input; and

a content analyzer having a content input coupled to the mixer output, a content output, and a content processor for calculating and providing to the content output a content signal representative of a calculated value of the source signal.

2. The recognition system as in claim 1, wherein the content processor further calculates and provides to the content output a content probability signal representative of a confidence level in the calculated value of the source signal.

3. The recognition system as in claim 2, wherein the input means includes at least one microphone.

4. A method for converting a source signal to a symbolic signal comprising the steps of:

obtaining at least one input signal from the source signal;

calculating a probability that the source signal originates from one of at least one predetermined directions;

generating a noise signal inversely proportional to the calculated probability;

mixing the at least one input signal with the generated noise signal; and

generating from the mixed signal a content signal representative of a probable value of the source signal.