Voice web search
View Patent ↗A search system will receive a voice query and use speech recognition with a predefined vocabulary to generate a textual transcription of the voice query. Queries are sent to a text search engine, retrieving multiple web page results for each of these initial text queries. The collection of the keywords is extracted from the resulting web pages and is phonetically indexed to form a voice query dependent and phonetically searchable index database. Finally, a phonetically-based voice search engine is used to search the original voice query against the voice query dependent and phonetically searchable index database to find the keywords and/or key phrases that best match what was originally spoken. The keywords and/or key phrases that best match what was originally spoken are then used as a final text query for a search engine. Search results from the final text query are then presented to the user.
1. A computer-implemented method, comprising:
receiving, at a computing system having one or more processors, a speech input corresponding to a web-based search query;
obtaining, by the computing system, a plurality of possible transcriptions of the speech input;
selecting, by the computing system, a plurality of transcriptions having highest confidence scores from the plurality of possible transcriptions to obtain a set of selected transcriptions;
obtaining, by the computing system, first web-based search results for each transcription of the set of selected transcriptions;
obtaining, by the computing system, extracted keywords from a web page associated with each first web-based search result;
performing, by the computing system, a voice-to-text search of the first web-based search results using the speech input and the extracted keywords to obtain second web-based search results; and
outputting, from the computing system, the second web-based search results.
2. The computer-implemented method of claim 1 , wherein obtaining the second web-based search results includes:
obtaining, by the computing system, a search string as a result of the voice-to-text search of the first web-based search results using the speech input and the extracted keywords; and
initiating, by the computing system, a web-based search using the search string to obtain (i) the second web-based search results and (ii) a ranking of the second web-based search results indicative of a relative importance of each of the second web-based search results.
3. The computer-implemented method of claim 1 , wherein selecting the set of selected transcriptions includes:
obtaining, by the computing system, the confidence score for each possible transcription, the confidence score being indicative of a likelihood that a specific possible transcription is a correct transcription of the speech input; and
selecting, by the computing system, ones of the plurality of possible transcriptions having the highest confidence scores to obtain the set of selected transcriptions.
4. The computer-implemented method of claim 1 , wherein the extracted keywords are derived from metadata for webpages corresponding to the first web-based search results.
5. The computer-implemented method of claim 4 , wherein a specific metadata includes at least one keyword associated with a corresponding webpage.
6. The computer-implemented method of claim 1 , wherein the computing system includes at least one remote server.
7. A computing system having one or more processors configured to perform operations comprising:
receiving a speech input corresponding to a web-based search query;
obtaining a plurality of possible transcriptions of the speech input;
selecting a plurality of transcriptions having highest confidence scores from the plurality of possible transcriptions to obtain a set of selected transcriptions;
obtaining first web-based search results for each transcription of the set of selected transcriptions;
obtaining extracted web content from a web page associated with each first web-based search result;
performing a voice-to-text search of the first web-based search results using the speech input and the extracted web content to obtain second web-based search results; and
outputting the second web-based search results.
8. The computing system of claim 7 , wherein obtaining the second web-based search results includes:
obtaining a search string as a result of the voice-to-text search of the first web-based search results using the speech input and the extracted keywords; and
initiating, by the computing system, a web-based search using the search string to obtain (i) the second web-based search results and (ii) a ranking of the second web-based search results indicative of a relative importance of each of the second web-based search results.
9. The computing system of claim 7 , wherein selecting the set of selected transcriptions includes:
obtaining the confidence score for each possible transcription, the confidence score being indicative of a likelihood that a specific possible transcription is a correct transcription of the speech input; and
selecting ones of the plurality of possible transcriptions having the highest confidence scores to obtain the set of selected transcriptions.
10. The computing system of claim 7 , wherein the extracted keywords are derived from metadata for webpages corresponding to the first web-based search results.
11. The computing system of claim 10 , wherein a specific metadata includes at least one keyword associated with a corresponding webpage.
12. The computing system of claim 7 , wherein the computing system includes at least one remote server.
13. A non-transitory computer-readable medium having instructions stored thereon that, when executed by one or more processors of a computing system, cause the computing system to perform operations comprising:
receiving a speech input corresponding to a web-based search query;
obtaining a plurality of possible transcriptions of the speech input;
selecting most-likely transcriptions from the plurality of possible transcriptions to obtain a set of selected transcriptions;
obtaining first web-based search results for each transcription of the set of selected transcriptions;
obtaining extracted web content from each first web-based search result;
performing a voice-to-text search of the first web-based search results using the speech input and the extracted web content to obtain second web-based search results; and
outputting the second web-based search results.
14. The computer-readable medium of claim 13 , wherein obtaining the second web-based search results includes:
obtaining a search string as a result of the voice-to-text search of the first web-based search results using the speech input and the extracted keywords; and
initiating, by the computing system, a web-based search using the search string to obtain (i) the second web-based search results and (ii) a ranking of the second web-based search results indicative of a relative importance of each of the second web-based search results.
15. The computer-readable medium of claim 13 , wherein selecting the most-likely transcriptions includes:
obtaining a confidence score for each possible transcription indicative of a likelihood that the possible transcription is a correct transcription of the speech input; and
selecting ones of the plurality of possible transcriptions having the highest confidence scores to obtain the set of selected transcriptions.
16. The computer-readable medium of claim 13 , wherein the extracted keywords are derived from metadata for webpages corresponding to the first web-based search results, and wherein a specific metadata includes at least one keyword associated with a corresponding webpage.
17. The computer-readable medium of claim 13 , wherein the computing system includes at least one remote server.