Keyword detection in continuous speech using continuous asynchronous correlation
First Claim
1. A system for asynchronously detecting one or more keywords in continuous speech, said system comprising:
- processor means responsive to a continuous speech signal for continuously developing a plurality of spectral estimates therefrom;
generator means responsive to the plurality of spectral estimates for periodically generating digitized spectral parameters therefrom;
first memory means for prestoring predetermined ones of dyad and triad subelement signals associated with one or more preselected keywords;
correlation means for asynchronously correlating each of the prestored dyad and triad subelement signals of one or more preselected keywords with continuous sequences of the digitized spectral parameters to asynchronously produce a plurality of continuous correlation data signals; and
a decision means for each desired keyword, each of said decision means being responsive to associated ones of the correlation data signals for developing an occurrence decision output whenever an associated keyword has been detected.
0 Assignments
0 Petitions
Accused Products
Abstract
A system for detecting one or more keywords in continuous speech is disclosed wherein a speech processor extracts a plurality of analog speech parameters from the input continuous speech, an STV (Spectral Time Vector) generator circuit selectively converts the plurality of analog speech parameters into digitized spectral parameters, an asynchronous correlation circuit continuously correlates sequences of the digitized spectral parameters with stored reference mask templates representative of subelements (dyads or triads) of one or more desired keywords in order to produce continuous correlation data, and a decision function circuit is responsive to the correlation data for developing an occurrence decision output for each desired keyword that is detected.
-
Citations
35 Claims
-
1. A system for asynchronously detecting one or more keywords in continuous speech, said system comprising:
-
processor means responsive to a continuous speech signal for continuously developing a plurality of spectral estimates therefrom; generator means responsive to the plurality of spectral estimates for periodically generating digitized spectral parameters therefrom; first memory means for prestoring predetermined ones of dyad and triad subelement signals associated with one or more preselected keywords; correlation means for asynchronously correlating each of the prestored dyad and triad subelement signals of one or more preselected keywords with continuous sequences of the digitized spectral parameters to asynchronously produce a plurality of continuous correlation data signals; and a decision means for each desired keyword, each of said decision means being responsive to associated ones of the correlation data signals for developing an occurrence decision output whenever an associated keyword has been detected. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25)
-
-
26. A system for asynchronously detecting one or more keywords in continuous speech, said system comprising:
-
input means for separating an input continuous speech signal into a plurality of frequency components; parameter means for developing a plurality of spectral estimates in response to the plurality of frequency components; generator means for periodically generating digitized spectral parameters in response to the plurality of spectral estimates; first memory means for storing sequences of the periodically generated digitized spectral parameters; second memory means for prestoring preselected ones of dyad and triad subelements associated with at least one preselected keyword; means for asynchronously correlating each of the prestored dyad and triad subelements with a correspondingly long sequence of digitized spectral parameters during a read mode of operation to asynchronously develop a corresponding correlation data signal for each prestored subelement; and a decision means for each desired keyword, each of said decision means being responsive to associated ones of the correlation data signals for developing an occurrence decision output whenever an associated keyword is detected. - View Dependent Claims (27, 28)
-
-
29. A system for asynchronously detecting one or more keywords in continuous speech, said system comprising:
-
first means responsive to a continuous speech signal for periodically generating a plurality of digitized spectral parameters therefrom; means for prestoring predetermined ones of dyad and triad subelement signals associated with one or more preselected element signals associated with one or more preselected keywords; correlation means for asynchrously correlating each of the prestored dyad and triad subelement signals of one or more preselected keywords with continuous sequences of the digitized spectral parameters to asynchronously produce a plurality of continuous correlation data signals; and a decision means for each desired keyword, each of said decision means being responsive to associated ones of the correlation data signals for developing an occurrence decision output whenever an associated keyword has been detected.
-
-
30. A system for asynchronously detecting keywords in continuous speech, said system comprising:
-
means for generating a continuous sequence of digitized spectral parameters from the continuous speech; means for prestoring predetermined ones of dyad and triad subelement signals associated with one or more preselected keywords; means for asynchronously correlating each of the prestored dyad and triad subelements of the preselected keywords with the continuous sequence of digitized spectral parameters to asynchronously produce a continuous sequence of respective correlation data signals; and means for developing an output signal indicative of the detection of a keyword whenever an associated sequence of correlation data signals develops a function signal having an amplitude greater than a predetermined threshold level. - View Dependent Claims (31)
-
-
32. A machine method for asynchronously detecting at least one preselected keyword in an input continuous speech signal comprising the steps of:
-
generating periodically a plurality of digitized spectral parameters from the input continuous speech signal; correlating asynchronously each of a plurality of preselected ones of dyad and triad subelement signals associated with said at least one preselected keyword with continuous sequences of pluralities of digitized spectral parameters to asynchronously produce a plurality of continuous correlation data signals; and developing an occurrence decision output signal for each associated keyword that is detected in an associated sequence of correlation data signals.
-
-
33. A machine method for asynchronously detecting at least one preselected keyword in an input continuous speech signal comprising the steps of:
-
producing a plurality of spectral estimates continuously from the input continuous speech signal; generating digitized spectral parameters periodically from the plurality of spectral estimates; correlating asynchronously each of a plurality of preselected ones of dyad and triad subelements associated with said at least one preselected keyword with continuous sequences of the digitized spectral parameters to asynchronously produce a plurality of continuous correlation data signals; and developing an occurrence decision output signal for each associated keyword that is detected in an associated sequence of correlation data signals.
-
-
34. A machine method for asynchronously detecting preselected keywords in continuous speech comprising the steps of:
-
producing a continuous sequence of digitized spectral parameters from the continuous speech; correlating asynchronously each of a plurality of preselected ones of dyad and triad subelements associated with the preselected keywords with the digitized spectral parameters to asynchronously develop respective correlation data signals; and developing an output signal indicative of the detection of a keyword whenever an associated sequence of correlation data signals develops a function signal having an amplitude greater than a predetermined threshold level. - View Dependent Claims (35)
-
Specification