Word boundary detector for speech recognition equipment

• US 4,032,710 A
• Filed: 03/10/1975
• Issued: 06/28/1977
• Est. Priority Date: 03/10/1975
• Status: Expired due to Term

• Current Assignee: Siemens Corporate Research & Support, Inc. (Siemens AG)
• Original Assignee: Threshold Technology, Inc.
• Inventors: Cox, Robert B., Herscher, Marvin B., Martin, Thomas B.
• Primary Examiner(s): Cooper, William C.
• Assistant Examiner(s): Kemeny, E. S.
• Application Number:
  US05/556,633
• Time in Patent Office: 841 Days
• Field of Search: 179/1 SA, 179/1 SD, 179/1 SE, 179/1 SC
• US Class Current: 704/253
• CPC Class Codes: G10L 15/00 Speech recognition G10L17/00 takes precedence
• CPC Class Codes: G10L 25/87 Detection of discrete points within a voice signal

Abstract

The present invention pertains to an apparatus which receives acoustic input, the input including words spoken in isolation, finds the word boundary instants at which a word begins and ends, and performs recognition functions on the words. A feature of the invention is the compensation for breath noise after the true end of a word, using a variable backup of the estimated word end. The apparatus includes means for generating feature signals indicative of feature characteristics in the received input and further includes means for comparing the feature signals which occurred during determined time boundaries with stored features corresponding to words in a vocabulary. The invention is directed to an improved system for detecting word boundaries which includes a means responsive to the input for generating a first feature signal indicative of the substantially continuing presence of speech-like sounds which meet a first selection criterion. Means are provided for storing the feature signals which occur during the presence of this first feature signal. Further means, responsive to the input, are provided for generating a second feature signal indicative of the presence of speech-like sounds which meet a second selection criterion, this second selection criterion being more restrictive than the first selection criterion. Means are also provided for determining the last occurrence of the second feature signal among the stored feature signals. The end boundary of an input spoken word is determined as a function of this last occurrence.

Claims

1. In an apparatus which receives acoustic input, said input including words spoken in isolation, and performs recognition functions on said words, said apparatus including means for generating feature signals indicative of feature characteristics in the received input and means for comparing the feature signals which occur during determined time boundaries with stored features corresponding to words in a vocabulary;

an improved system for detecting word boundaries, comprising;

(a) means responsive to said input for generating a first feature signal indicative of the presence of speech-like sounds which meet a first selection criterion;

(b) means for storing the feature signals which occur during the presence of said first feature signal;

(c) means responsive to said input for generating a second feature signal indicative of the presence of speech-like sounds which meet a second selection criterion; and

(d) means for determining the substantially last occurrence of said second feature signal among the stored feature signals, the end boundary of an input spoken word being determined as a function of said last occurrence.

2. The system as defined by claim 1 wherein said second selection criterion is more restrictive than said first selection criterion.

3. The system as defined by claim 2 wherein said first feature signal is provided with a predetermined delay in its turn-off characteristic.

4. The system as defined by claim 1 wherein said means for generating said second feature signal includes means responsive to said input for generating an indication of the presence of a voiced phonetic characteristic in said input.

5. The system as defined by claim 1 wherein said means for generating said second feature signal includes means responsive to said input for generating an indication of the presence of an unvoiced noise-like consonant characteristic in said input.

6. The system as defined by claim 1 wherein said means for generating said second feature signal includes means responsive to said input for generating an indication of the presence of a voiced phonetic characteristic or an unvoiced noise-like consonant characteristic in said input.

7. The system as defined by claim 3 wherein said means for generating said second feature signal includes means responsive to said input for generating an indication of the presence of a voiced phonetic characteristic or an unvoiced noise-like consonant characteristic in said input.

8. The system as defined by claim 6 wherein said means for generating said second feature signal further includes means responsive to said input for generating an indication of the presence of a slowly decaying speech energy characteristic in said input.

9. In an apparatus which receives acoustic input, said input including words spoken in isolation, and performs recognition functions on said words, said apparatus including means for generating feature signals indicative of feature characteristics in the received input, and means for comparing the feature signals which occur during determined time boundaries with stored features corresponding to words in a vocabulary;

an improved system for detecting word boundaries, comprising;

(a) means responsive to said input for generating a first feature signal indicative of the presence of speech-like sounds which meet a predetermined energy threshold criterion;

(b) means for storing the feature signals which occur during the presence of said first feature signal;

(c) means responsive to said input for generating a third feature signal indicative of the presence of a voiced phonetic characteristic in said input;

(d) means responsive to said input for generating a fourth feature signal indicative of the presence of an unvoiced noise-like consonant in said input;

(e) means for generating a second feature signal as a function of said third and fourth feature signals; and

(f) means for determining the substantially last occurrence of said second feature signal among the stored feature signals, the end boundary of an input spoken word being a function of said last occurrence.

10. A system as defined by claim 9 wherein said first feature signal has a predetermined delay in its turnoff characteristic.

11. In conjunction with an apparatus which receives acoustic input that includes spoken words in isolation and performs recognition functions on said words, the apparatus generating signals indicative of feature characteristics in the received input and comparing signals which occur during determined time boundaries with stored features corresponding to words in a vocabulary;

a method for detecting word boundaries, comprising the steps of;

(a) generating a first feature signal indicative of the presence of speech-like sounds which meet a first selection criterion;

(b) storing the feature signals which occur during the presence of said first feature signal;

(c) generating a second feature signal indicative of the presence of speech-like sounds which meet a second more restrictive selection criterion; and

(d) determining the substantially last occurrence of the second feature signal among the stored feature signals, the end boundary of an input spoken word being a function of said last occurrence.