Voice recognition system

• US 5,003,603 A
• Filed: 08/20/1984
• Issued: 03/26/1991
• Est. Priority Date: 08/20/1984
• Status: Expired due to Term

• Current Assignee: Voiceit, Ltd.
• Original Assignee: Franz Kavan, Gus Searcy
• Inventors: Searcy, Gus, Kavan, Franz
• Primary Examiner(s): KEMENY, EMANUEL
• Application Number:
  US06/642,299
• Time in Patent Office: 2,409 Days
• Field of Search: 381/41-43, 381/51, 364/513, 364/513.5
• US Class Current: 704/253
• CPC Class Codes: G10L 15/02 Feature extraction for speech recognition; Selection of recognition unit
• CPC Class Codes: G10L 15/10 using distance or distortion measures between unknown speech and reference templates
• CPC Class Codes: G10L 25/09 the extracted parameters being zero crossing rates

Abstract

A method and apparatus are disclosed for recognizing spoken commands uttered by a user and for generating responsive control signals once the command is recognized. In accordance with this disclosure the audio signal is converted into a series of count bytes representing the time between the audio signal zero crossings, and all the count bytes of the full command are then segmented into equal temporal groups histogram and sorted within each segment into a set of frequency class intervals which are based on a computation of substantially equal byte activity in all the words comprising the command lexicon. In this manner, lower and higher frequency groups are selected for equal significance. The uttered words are then compared against stored words similarly transformed according to segment and frequency interval and if the comparison conditions are satisfied the command is executed; if not, an indication is provided to the user to repeat the command.

Claims

1. A speech recognition system comprising:

an audio sensing circuit conformed to provide a sequence of pulses each of a length corresponding to the length of an audio wave;

buffer memory means conformed to receive and store a sequence of count codes each corresponding to the length of a respective one of said pulses; and

sorting means operatively connected to receive said count codes from said buffer memory means and conformed to sort said count codes into preselected class intervals of count groups,

said class intervals of count groups being selected to include a substantially equal number of pulses produced in the course of speech.

2. Apparatus according to claim 1 wherein:

said audio sensing circuit includes a bandpass filter conformed to pass the frequency bands associated with human speech.

3. A method for recognizing words uttered by a user comprising the steps of:

converting the audio wave forms characterizing said words into pulses of a length corresponding to said waves;

producing a serial stream of count codes indicative of the duration of each said pulses;

sorting said count codes in accordance with the count thereof and accumulating the number of occurrences of a count code according to preselected count intervals; and

storing said number of count codes in each said interval;

said count intervals are preselected to provide a substantially equal count number accumulated for all said words uttered by the user.

4. A method according to claim 3 wherein said step of producing a serial stream further comprises the steps of:

accumulating said count codes to produce a sum thereof;

dividing said sum by a predetermined integer number to produce segments of count codes; and

selecting said serial stream of count codes from each segment.

5. A method according to claim 3 further comprising:

comparing the count codes of a word uttered by the user and selected according to said segments and sorted according to said intervals with said stored selected and sorted count codes.

6. Apparatus for providing a predetermined response signal uniquely associated with the reception of a particular word uttered by a user comprising:

a microphone deployed to sense the audio waves characterizing each said word for producing an electrical signal indicative thereof;

filtering means connected to receive said electrical signal for producing an output signal attenuated in selected frequency components thereof;

pulse generating means connected to said filtering means for producing a sequence of pulses each of a length corresponding to the duration between zero amplitude crossings of said output signal; and

signal processing means connected to receive said sequence of pulses from said pulse generating means, said signal processing means including;

(a) pulse length counting means connected to receive each said pulse for producing a count code byte of a code corresponding to the length of said pulse at the completion of each said pulse;

(b) sorting means operatively connected to receive said count code bytes for sorting thereof into predetermined intervals of count code and for storing the number of bytes sorted in each said interval said count code intervals being each selected to include a substantially equal number of count bytes accumulated for all said byte number patterns;

(c) memory means conformed to store reference byte number patterns in accordance with said intervals; and

(d) comparison means connected to receive said byte numbers sorted in accordance with said intervals in said sorting means and said reference byte number patterns for producing said response signal upon a coincidence therebetween within a predetermined tolerance.

7. Apparatus according to claim 6 wherein:

said filtering means is selected to pass frequencies between 300 and 3000 cycles per second and to attenuate all other frequencies.

8. Apparatus according to claim 7 wherein:

said count code intervals are each selected to include a substantially equal number of count code bytes accumulated for all said byte number patterns.

9. Apparatus according to claim 8 wherein:

said filtering means is selected to pass frequencies between 300 and 3000 cycles per second and to attenuate all other frequencies.