Automatic language model update
First Claim
1. A computer-implemented method for generating a speech recognition model, comprising:
- accessing a baseline speech recognition model installed on a server system;
obtaining, at the server system, information from recent text search queries submitted by a plurality of users to a search system, wherein the information (i)specifies a frequency of occurrence for one or more words in the recent search queries and (ii) identifies times when instances of the one or more words were submitted to the search system in recent search queries; and
modifying, by the server system, the speech recognition model to revise probabilities of a portion of a sound occurrence based on the information obtained from the recent text search queries,wherein the modification weights influences of instances of the one or more words on the revised probabilities based on times when the one or more words were submitted to the search system in recent search queries.
2 Assignments
0 Petitions
Accused Products
Abstract
A method for generating a speech recognition model includes accessing a baseline speech recognition model, obtaining information related to recent language usage from search queries, and modifying the speech recognition model to revise probabilities of a portion of a sound occurrence based on the information. The portion of a sound may include a word. Also, a method for generating a speech recognition model, includes receiving at a search engine from a remote device an audio recording and a transcript that substantially represents at least a portion of the audio recording, synchronizing the transcript with the audio recording, extracting one or more letters from the transcript and extracting the associated pronunciation of the one or more letters from the audio recording, and generating a dictionary entry in a pronunciation dictionary.
-
Citations
26 Claims
-
1. A computer-implemented method for generating a speech recognition model, comprising:
-
accessing a baseline speech recognition model installed on a server system; obtaining, at the server system, information from recent text search queries submitted by a plurality of users to a search system, wherein the information (i)specifies a frequency of occurrence for one or more words in the recent search queries and (ii) identifies times when instances of the one or more words were submitted to the search system in recent search queries; and modifying, by the server system, the speech recognition model to revise probabilities of a portion of a sound occurrence based on the information obtained from the recent text search queries, wherein the modification weights influences of instances of the one or more words on the revised probabilities based on times when the one or more words were submitted to the search system in recent search queries. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14)
-
-
15. A method for generating a speech recognition model, comprising:
-
receiving at a search engine from a remote device an audio recording and a transcript that substantially represents at least a portion of the audio recording; synchronizing the transcript with the audio recording; extracting one or more letters from the transcript and extracting an associated pronunciation of the one or more letters from the audio recording; generating a dictionary entry in a pronunciation dictionary; and assigning probability weightings to the dictionary entries based on when the audio recording was generated. - View Dependent Claims (16, 17, 18, 19, 20, 21, 22, 23)
-
-
24. A system for updating a language model, comprising:
-
a request processor to receive text search terms that are from recent text search queries submitted by a plurality of users; an extractor for obtaining information from the recent text search queries, wherein the information (i) specifies a frequency of occurrence for one or more words in the recent text search queries and (ii) identifies when instances of the one or more words were received as text queries; and an updater for modifying a language model to revise probabilities of a word occurrence based on the information obtained from the recent text search queries, wherein the modification weights the influence of instances of the one or more words on the revised probabilities based on when the one or more words were received as text queries.
-
-
25. A computer implemented method for transmitting verbal terms, comprising:
-
transmitting text search terms from a remote device to a server system, wherein the server system; generates word occurrence data that specifies a probability of occurrence of a word, and is associated with the text search terms and a frequency of occurrence of recent text search queries submitted by a plurality of users to the server system, and identifies times when the recent text search queries were transmitted to the server system, and modifies a language model based on the word occurrence data, wherein the modification weights probabilities for the word occurrence data based on times when the one or more words were transmitted as text queries. - View Dependent Claims (26)
-
Specification