System and methods for improving language pronunciation

• US 9,076,347 B2
• Filed: 03/14/2013
• Issued: 07/07/2015
• Est. Priority Date: 03/14/2013
• Status: Active Grant

• Current Assignee: Better Accent, LLC
• Original Assignee: Better Accent, LLC
• Inventors: Komissarchik, Julia, Komissarchik, Edward
• Primary Examiner(s): Chawan, Vijay B
• Application Number:
  US13/828,843
• Publication Number:
  US 20140278421A1
• Time in Patent Office: 845 Days
• Field of Search: 704/254, 704/251, 704/260, 704/270, 704/2, 704/8, 704/257, 434/167, 434/156, 434/157, 434/185, 434/169, 381/98, 715/264
• US Class Current: 1/1
• CPC Class Codes: G09B 19/06 Foreign languages with audible presentation of material to be studied G09B5/04
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence
• CPC Class Codes: G10L 25/48 specially adapted for particular use

Abstract

A system and methods for analyzing pronunciation, detecting errors and providing automatic feedback to help non-native speakers improve pronunciation of a foreign language is provided that employs publicly available, high accuracy third-party automatic speech recognizers available via the Internet to analyze and identify mispronunciations.

Claims

1. A system for identifying a mispronounced phrase spoken by a user of a non-native language, comprising:

a curriculum program module comprising a plurality of tracks, each track corresponding to a selected set phrases;

a speech recognition system that analyzes an utterance spoken by the user, the utterance corresponding to a phrase of the track, the speech recognition system returning a ranked list of two or more recognized phrases;

a pronunciation analysis module that;

compares each of the ranked list of two or more recognized phrases to a series of mispronunciations corresponding to the selected phrase;

matches one of the ranked list of two or more recognized phrases to a mispronunciation; and

identifies guidance to correct the matched mispronunciation based upon the match of the one of the ranked list of two or more recognized phrases to a mispronunciation; and

a human-machine interface that enables the user to select a curriculum, track a phrase, prompts the user to speak the utterance; and

displays to the user the guidance identified by the pronunciation analysis module.

2. The system of claim 1, wherein the plurality of tracks are stored in a repository accessible via the Internet.

3. The system of claim 1, wherein the plurality of tracks are arranged by topic or theme.

4. The system of claim 1, further comprising a performance data analytics module that stores a first indication corresponding to the user'"'"'s progress through a selected one of the plurality of tracks.

5. The system of claim 4, wherein the performance data analytics module stores a second indication corresponding to the user'"'"'s progress through the plurality of tracks.

6. The system of claim 1, wherein the speech recognition system is accessible via the Internet.

7. The system of claim 6, wherein the speech recognition system comprises a publicly-available third-party speech recognition system.

8. The system of claim 1, wherein the series of mispronunciations corresponding to the selected phrase is logically arranged in a star configuration, comprising a central node corresponding to the selected phrase, and a ray corresponding to a mispronunciation of the selected phrase, wherein the phonetic transcription of the ray differs by no more than three elements from the phonetic transcription of the central node.

9. The system of claim 1, further comprising a mispronunciation repository accessible via the Internet, wherein the mispronunciation repository stores mispronunciations corresponding to phrases in the plurality of tracks.

10. The system of claim 1 wherein the human-machine interface is configured to operate on a mobile device.

19. The system of claim 1, wherein the pronunciation analysis module further configured to filter one or more phrases from the ranked list of two or more recognized phrases based upon a confidence level threshold.

20. The system of claim 1, wherein the pronunciation analysis module further configured to determine that the first n matches of the ranked list of two or more recognized phrases do not match the phrase of the track, wherein n is greater than one, and wherein the matched one of the ranked list is the highest ranking phrases in the ranked list.

11. A method for identifying a mispronounced word spoken by a user of a non-native language, comprising:

presenting to the user a curriculum comprising a plurality of tracks, each track corresponding to a selected set of phrases;

prompting the user to create an utterance corresponding to a phrase selected from among the plurality of tracks;

analyzing, one or more electronic processors, the utterance using a speech recognition system, the speech recognition system returning a ranked list of two or more recognized phrases;

comparing, the one or more electronic processors, each of the ranked list of two or more recognized phrases to a series of mispronunciations corresponding to the phrase;

matching, using the one or more electronic processors, one of the ranked list of two or more recognized phrases to a mispronunciation;

identifying, using the one or more electronic processors, guidance for the user to correct the matched mispronunciation based upon the match of the one of the ranked list of two or more recognized phrases to a mispronunciation; and

displaying the guidance to the user.

12. The method of claim 11, further comprising arranging the plurality of tracks by topic or theme.

13. The method of claim 11, further comprising storing a first indication corresponding to the user'"'"'s progress through a selected one of the plurality of tracks.

14. The method of claim 13, further comprising storing a second indication corresponding to the user'"'"'s progress through the plurality of tracks.

15. The method of claim 11, further comprising accessing the speech recognition system via the Internet.

16. The method of claim 15, wherein accessing the speech recognition system via the Internet comprises accessing a publicly-available third-party speech recognition system.

17. The method of claim 11, further comprising logically arranging the series of mispronunciations corresponding to the phrase in a star configuration comprising a central node corresponding to the phrase, and a ray corresponding to a mispronunciation of the phrase, wherein the phonetic transcription of the ray differs by no more than three elements from the phonetic transcription of the central node.

18. The method of claim 11, further comprising storing mispronunciations corresponding to phrases in the plurality of tracks in a repository accessible via the Internet.