Mobile dictation correction user interface
View Patent ↗A method of speech recognition is described for use with mobile devices. A portion of an initial speech recognition result is presented on the mobile device including a set of general alternate recognition hypotheses associated with the portion of the speech recognition result. A key input representative of one or more associated letters is received from the user. The user is provided with a set of restricted alternate recognition hypotheses starting with the one or more letters associated with the key input. Then a user selection is accepted of one of the restricted alternate recognition hypotheses to represent a corrected speech recognition result.
1. A method of speech recognition on a mobile device comprising:
presenting with the mobile device for a speech input containing a plurality of spoken words, a recognized word display representing one or more most likely recognition hypotheses corresponding to a current one of the spoken words;
performing a recognition verification process wherein recognition of each spoken word is veified by a user input action. the verification process including either:
receiving from the user a key input verifying one of the displayed recognition hypotheses as correct, or
receiving from the user:
i. a first key input representative of one or more associated letters which limits the recognition hypotheses presented in the recognized word display to a limited set of one or more recognition hypotheses starting with the one or more letters associated with the key input, and
ii. accepting a second key input verifying one of the recognition hypotheses presented in the recognized word display as a corrected speech recognition result; and
repeating the recognition verification process for a next word in the speech input until all the spoken words have been processed.
2. The method according to claim 1 , further comprising:
providing the verified speech recognition results to a text application.
3. The method according to claim 2 , further comprising:
providing an currently amended audio file with the verified speech recognition results.
4. The method according to claim 3 , wherein the currently amended audio file is provided using a URL pointer to a storage location of the currently amended audio file.
5. The method according to claim 1 , wherein a word lattice contains the most likely recognition hypotheses.
6. The method according to claim 1 , wherein a recognition sausage contains the most likely recognition hypotheses.
7. The method according to claim 1 , wherein the recognition hypotheses are derived via a phone to letter algorithm.
8. A speech recognition user correction interface for a mobile device comprising:
presenting means for presenting with the mobile device for a speech input containing a plurality of spoken words, a recognized word display representing one or more most likely recognition hypotheses corresponding to a current one of the spoken words;
recognition verification process means for performing a recognition verification process wherein recognition of each spoken word is verified by a user input action, the verification process including either:
receiving from the user a key input verifying one of the displayed recognition hypotheses as correct, or
receiving from the user:
i. a first key input representative of one or more associated letters which limits the recognition hypotheses presented in the recognized word display to a limited set of one or more recognition hypotheses starting with the one or more letters associated with the key input, and
ii. a second key input verifying one of the recognition hypotheses presented in the recognized word display as a corrected speech recognition result; and
repeating means for repeating the recognition verification process means for a next word in the speech input until all the spoken words have been processed.
9. The interface according to claim 8 , further comprising:
result means for providing the verified speech recognition results to a text application.
10. The interface according to claim 9 , further comprising:
audio means for providing an currently amended audio file with the verified speech recognition results.
11. The interface according to claim 10 , wherein the audio means provides the currently amended audio file using a URL pointer to a storage location of the currently amended audio file.
12. The interface according to claim 8 , wherein a word lattice contains the most likely recognition hypotheses.
13. The interface according to claim 8 , wherein a recognition sausage contains the most likely recognition hypotheses.
14. The interface according to claim 8 , wherein the recognition hypotheses are derived via a phone to letter algorithm.
15. A mobile user device comprising:
a user verification interface including:
presenting means for presenting with the mobile device for a speech input containing a plurality of spoken words, a word display representing one or more most likely recognition hypotheses corresponding to a current one of the spoken words;
recognition verification process means for performing a recognition verification process wherein recognition of each spoken word is verified by a user input action. the verification process including either:
receiving from the user a key input verifying one of the displayed recognition hypotheses as correct, or
receiving from the user:
a first key input representative of one or more associated letters which limits the recognition hypotheses presented in the recognized word display to a limited set of one or more recognition hypotheses starting with the one or more letters associated with the key input;
ii. a second key input verifying one of the recognition hypotheses presented in the recognized word display as a corrected speech recognition result; and
repeating means for repeating the recognition verification process means for a next word in the speech input until all the spoken words have been processed.
16. The device according to claim 15 , further comprising:
result means for providing the verified speech recognition results to a text application.
17. The device according to claim 16 , further comprising:
audio means for providing an currently amended audio file with the verified speech recognition results.
18. The device according to claim 17 , wherein the audio means provides the currently amended audio file using a URL pointer to a storage location of the currently amended audio file.
19. The device according to claim 15 , wherein a word lattice contains the most likely recognition hypotheses.
20. The device according to claim 15 , wherein a recognition sausage contains the most likely recognition hypotheses.
21. The device according to claim 15 , wherein the recognition hypotheses are derived via a phone to letter algorithm.