Google was granted a patent today on a voice interface for searches.
There are a number of potential issues with understanding speech when trying to perform searches by voice, which are described by the patent filing.
With most speech recognition technology, there can be high error rates when the vocabulary used is large, and the amount of dialogue used is small. Those applications often need to be trained to recognize unique vocal inflections from speakers. A search query is often limited to a handful of words or less.
Voice interfaces for search engines currently limit that problem by keeping the scope of communication small, asking the searcher to choose from a limited number of categories, and drilling down to smaller categories while providing limited sets of choices.
This can cause voice search to be slow, and limited to pre-chosen categories from the search engine.
This patent application is an attempt to address those issues, by trying to decide which are the most likely words comprising the query, and searching for those.
Voice interface for a search engine
Inventors: Alexander Mark Franz, Monika H. Henzinger, Sergey Brin, and Brian Christopher Milch
Assigned to Google Inc.
United States Patent 7,027,987
Granted April 11, 2006
Filed: February 7, 2001
A system provides search results from a voice search query. The system receives a voice search query from a user, derives one or more recognition hypotheses, each being associated with a weight, from the voice search query, and constructs a weighted boolean query using the recognition hypotheses. The system then provides the weighted boolean query to a search system and provides the results of the search system to a user.
The patent points to a number of other patents that are noted as references cited. It’s an interesting list:
- System and method for fragmentation mapping (4771384) granted September, 1988
- Speech recognition system with efficient storage and rapid assembly of phonological graphs (4980918) granted December, 1990
- Information retrieval system and method for displaying and ordering information based on query element contribution (5826260) granted October, 1998
- Interactive searching (6112172) granted August, 2000
- Information retrieval system and method that generates weighted comparison results to analyze the degree of dissimilarity between a reference corpus and a candidate document (6167398) granted December, 2000
- Multimedia document retrieval by application of multimedia queries to a unified index of multimedia data for a plurality of multimedia data types (6243713) granted June, 2001
- Data mining using cyclic association rules (6278998) granted August, 2001
- Compound words in speech recognition systems (6349282) granted February, 2002
- Methods and apparatus for forming compound words for use in a continuous speech recognition system (6385579) granted May, 2002
- Compound word recognition (6393399) granted May, 2002
- Information retrieval and speech recognition based on language models (6418431) granted July, 2002
- Object interactive user interface using speech recognition and natural language processing (6434524) granted August, 2002
- Minimization of search network in speech recognition (6456970) granted September, 2002
- Network interactive user interface using speech recognition and natural language processing (6532444) granted March, 2003
- System and method for incorporating concept-based retrieval within boolean search engines (6745161) granted June, 2004