| 摘要 |
<p>A method and apparatus for voice searching is disclosed. The method may include receiving a search query (5150) from a user of a mobile communication device (120), converting speech parts in the search query into linguistic representations, generating a search phoneme lattice (5200) based on the linguistic representations, extracting query features (5250) from the generated search phoneme lattice, generating coarse search feature vectors (5300) based on the extracted query features, performing a coarse search (5350) using the generated coarse search feature vectors, transmitting the generated coarse search feature vectors (5400) to a remote voice search processing unit (130), receiving web indices (5450) from the remote voice search processing unit (130) based on the generated coarse search feature vectors, performing a fine search (5600) using the results of the coarse search, the remote web indices, and generated fine search feature vectors, and displaying the fine search results (5650) to the user.</p> |
