Abstract: A voice recognition method includes: representing a vocabulary translated into a Markov network; decoding by means of a Viterbi algorithm; and pruning explored solutions. The vocabulary is described in a tree made up of arcs and of nodes, between which transcriptions are defined that describe phonetic units used by a language model. The Markov network is constructed dynamically at least in part by means of Markov sub-units.

Abstract: A programmable data processing device for a digital TV set-top box comprises: a loading engine for receiving portions of code of a first type and/or data from a DSM-CC Carousel, a storage means for storing the portions received by the loading engine, an execution engine for executing an application embodied by the received portions; and a translating engine for translating the first type code into a native code of the execution engine. The translating engine is adapted to compile at least a certain one of said received portions into native code and to store the thus compiled portion in the storage means, and to interpret other portions of code, and the execution engine is adapted to process compiled code and interpreted code within a same application.