Method and apparatus for speech recognition

• US 10,140,974 B2
• Filed: 06/26/2015
• Issued: 11/27/2018
• Est. Priority Date: 12/29/2014
• Status: Active Grant

• Current Assignee: Samsung Electronics Co. Ltd.
• Original Assignee: Samsung Electronics Co. Ltd.
• Inventors: Hong, Seokjin, Choi, YoungSang
• Primary Examiner(s): Shah, Bharatkumar S
• Application Number:
  US14/751,654
• Publication Number:
  US 20160189710A1
• Time in Patent Office: 1,250 Days
• Field of Search: 704235
• CPC Class Codes: G10L 15/02 Feature extraction for speech recognition; Selection of recognition unit
• CPC Class Codes: G10L 2015/025 Phonemes, fenemes or fenones being the recognition units
• CPC Class Codes: G10L 2015/027 Syllables being the recognition units

Abstract

A method and apparatus for speech recognition are disclosed. The speech recognition apparatus includes a processor configured to process a received speech signal, generate a word sequence based on a phoneme sequence generated from the speech signal, generate a syllable sequence corresponding to a word element among words comprised in the word sequence based on the phoneme sequence, and determine a text corresponding to a recognition result of the speech signal based on the word sequence and the syllable sequence.

Claims

1. A speech recognition method, comprising:

generating, by a processor, a word sequence based on a phoneme sequence generated from a speech signal;

generating, by the processor, a syllable sequence based on the phoneme sequence, in response to a word element among words included in the word sequence having a lower recognition rate than a threshold value; and

determining, by the processor, a text corresponding to a recognition result of the speech signal based on the word sequence and the syllable sequence,

wherein the syllable sequence corresponds to the word element,

wherein the generating of the syllable sequence comprises generating the syllable sequence by decoding a portion corresponding to the word element, among phonemes included in the phoneme sequence, and

wherein the word sequence is generated by decoding the phoneme sequence and corresponds to at least the text corresponding to the recognition result of the speech signal.

2. The method of claim 1, wherein the determining of the text comprises determining the text by substituting the word element in the word sequence for the syllable sequence.

3. The method of claim 1, wherein the threshold value comprises a relative threshold value determined based on recognition rates of the words included in the word sequence.

4. The method of claim 1, wherein the generating of the syllable sequence based on the phoneme sequence comprises generating the syllable sequence by decoding a syllable unit using a syllable unit phonetic dictionary, in which a phoneme sequence comprising a syllable is modeled, and a syllable unit language model, in which a syllable sequence comprising a word is modeled.

5. The method of claim 1, wherein the syllable sequence comprises a word excluded from a word phonetic dictionary used to generate the word sequence based on the phoneme sequence.

9. A non-transitory computer-readable medium storing instructions that, when executed by the processor, cause the processor to perform the method of claim 1.

6. A speech recognition method, comprising:

generating, by a processor, a word sequence by decoding a phoneme sequence generated from a speech signal;

generating, by the processor, a syllable sequence corresponding to a word element, among words included in the word sequence, based on the phoneme sequence, in response to the word element comprising a lower recognition rate than a threshold value; and

determining, by the processor, a text corresponding to a recognition result of the speech signal based on the word sequence and the syllable sequence,

wherein the generating of the syllable sequence comprises generating the syllable sequence by decoding a portion corresponding to the word element among phonemes included in the phoneme sequence.

7. The method of claim 6, wherein the determining of the text comprises determining by substituting the word element in the word sequence for the syllable sequence.

8. The method of claim 6, wherein the threshold value is one of a threshold value from the user and a relative threshold value determined based on recognition rates of the words included in the word sequence.

10. A speech recognition apparatus, comprising:

a processor configured to

process a received speech signal,

generate a word sequence based on a phoneme sequence generated from the speech signal,

generate a syllable sequence corresponding to a word element among words included in the word sequence, based on decoding a portion corresponding to the word element among phonemes included in the phoneme sequence, in response to the word element having a lower recognition rate than a threshold value, and

determine a text corresponding to a recognition result of the speech signal based on the word sequence and the syllable sequence,

wherein the word sequence is generated by decoding the phoneme sequence and corresponds to at least the text corresponding to the recognition result of the speech signal.

11. The apparatus of claim 10, wherein the processor is configured to determine the text by substituting the word element in the word sequence for the syllable sequence.

12. The apparatus of claim 10, wherein the threshold value comprises a threshold value determined based on recognition rates of the words included in the word sequence.