| 摘要 |
<p>A method, apparatus and computer program product are provided for personalizing speech recognition data. A speech recognition model (SRM) that is adaptable by a user terminal based on user terminal dependent data may be received and adapted by a user terminal. A speaker dependent SRM may be refined on the user terminal and transmitted to a remote storage location, such as personalized speech recognition apparatus. The apparatus may cause transmission of SRMs to various user terminals, and may generate additional SRMs based on speaker dependent SRMs. Speaker dependent SRMs may be generated based on an individual, group of users, geographic location, dialect, or the like. SRMs may be based on hidden Markov Models,dynamic time warming models, neural networks, finite state transducers, or the like.</p> |
