Method and apparatus for recognizing a voice

1. A speech recognition method by a speech recognition device, the speech recognition method comprising:

buffering a spoken utterance;

extracting a standby wake-up word from the buffered spoken utterance by comparing the buffered spoken utterance to a preset wake-up word, the extracted standby wake-up word corresponding to the preset wake-up word;

analyzing a role of the extracted standby wake-up word in the buffered spoken utterance;

determining a speech intent in uttering the standby wake-up word in response to results of the analyzing the role of the extracted standby wake-up word such that the extracted standby wake-up word is determined as either a deliberate wake-up word or as an inadvertent wake-up word, the speech intent determined as utterance of the inadvertent wake-up word when the analyzed role of the extracted standby wake-up word is not a vocative case within the buffered spoken utterance, but is a sentence component including a subject or an object;

determining whether to execute a spoken sentence as a voice command included in the buffered spoken utterance, according to results of the determining the speech intent in uttering the standby wake-up word; and

processing the spoken sentence by:

activating a speech recognition function and executing the spoken sentence as the voice command in response to determining the extracted standby wake-up word as the deliberate wake-up word; and

not activating the speech recognition function in response to determining the extracted standby wake-up word as the inadvertent wake-up word.

2. The speech recognition method of claim 1 , wherein the buffering includes buffering the spoken utterance containing a pause between the standby wake-up word and the spoken sentence, the pause being shorter than a preset duration.

3. The speech recognition method of claim 1 , wherein the buffering further includes buffering the spoken utterance by varying a buffering volume or time in accordance with a condition under which the spoken utterance is inputted.

4. The speech recognition method of claim 1 , wherein the analyzing includes analyzing the role of the standby wake-up word, based on named-entity recognition or natural-language processing of the spoken utterance, the natural-language processing including one or more selected from among morphological segmentation, part-of-speech tagging, sentence breaking, and parsing.

5. The speech recognition method of claim 1 , further comprising, if the results of determining the speech intent in uttering the standby wake-up word indicate that the standby wake-up word is uttered as a deliberate wake-up word, isolating the spoken sentence without the deliberate wake-up word from the spoken utterance.

6. The speech recognition method of claim 5 , wherein the isolating includes at least one selected from:

isolating, from the spoken utterance, the spoken sentence preceding the deliberate wake-up word;

isolating, from the spoken utterance, the spoken sentence following the deliberate wake-up word;

isolating, from the spoken utterance, first and second spoken segments without the deliberate wake-up word positioned within the spoken sentence, the first spoken segment preceding the deliberate wake-up word and the second spoken segment following the deliberate wake-up word; and

isolating, from the spoken utterance containing two or more spoken sentences, the two or more spoken sentences without the deliberate wake-up word positioned between the two or more spoken sentences.

7. A computer program, stored in a non-transitory recording medium readable by a computer for permitting the speech recognition method of claim 1 to be executed by the computer.

8. A speech recognition device configured to recognize an inputted spoken utterance, the speech recognition device comprising:

a buffer configured to buffer the spoken utterance;

an extractor configured to extract, from the buffered spoken utterance, a standby wake-up word corresponding to a preset wake-up word by comparing the buffered spoken utterance to the preset wake-up word;

an analyzer configured to analyze a role of the extracted standby wake-up word within the buffered spoken utterance;

a determiner configured to determine a speech intent in uttering the standby wake-up word in response to results of the analyzing the role of the extracted standby wake-up word such that the extracted standby wake-up word is determined as either a deliberate wake-up word or as an inadvertent wake-up word, the speech intent determined as utterance of the inadvertent wake-up word when the analyzed role of the extracted standby wake-up word is not a vocative case within the buffered spoken utterance, but is a sentence component including a subject or an object; and

a controller configured to:

determine whether to execute a spoken sentence as a voice command included in the buffered spoken utterance, according to results of the determining the speech intent in uttering the standby wake-up word; and

process the spoken sentence by:

activating a speech recognition function and executing the spoken sentence as the voice command in response to determining the extracted standby wake-up word as the deliberate wake-up word; and

not activating the speech recognition function in response to determining the extracted standby wake-up word as the inadvertent wake-up word.

9. The speech recognition device of claim 8 , wherein the buffer is configured to buffer the spoken utterance containing a pause between the standby wake-up word and the spoken sentence, the pause being shorter than a preset duration.

10. The speech recognition device of claim 8 , further comprising a variator configured to vary a buffering volume or time of the buffer in accordance with a condition under which the spoken utterance is inputted.

11. The speech recognition device of claim 8 , wherein the analyzer is configured to analyze the role of the standby wake-up word, based on named-entity recognition or natural language processing of the spoken utterance, wherein the natural language processing performs one or more selected from among morphological segmentation, part-of-speech tagging, sentence breaking, and parsing.

12. The speech recognition device of claim 8 , further comprising an isolator, wherein, when the results of determining the speech intent in uttering the standby wake-up word indicate that the standby wake-up word is uttered as a deliberate wake-up word, the isolator is configured to isolate, from the spoken utterance, the spoken sentence without the deliberate wake-up word.

13. The speech recognition device of claim 12 , wherein the isolator performs at least one selected from:

isolating, from the spoken utterance, the spoken sentence preceding the deliberate wake-up word;

isolating, from the spoken utterance, the spoken sentence following the deliberate wake-up word;

isolating, from the spoken utterance, first and second spoken segments without the deliberate wake-up word, the first spoken segment preceding the deliberate wake-up word, the second spoken segment following the deliberate wake-up word, and the deliberate wake-up word being positioned between the first spoken segment and the second spoken segment; and

from the spoken utterance including two or more spoken sentences, isolating the two or more spoken sentences without the deliberate wake-up word positioned between the two or more spoken sentences.