Distributed pipelined parallel speech recognition system
A speech recognition circuit comprising a circuit for providing state identifiers which identify states corresponding to nodes or groups of adjacent nodes in a lexical tree, and for providing scores corresponding to said state identifiers, the lexical tree comprising a model of words. The circuit includes: a memory structure for receiving and storing state identifiers identified by a node identifier identifying a node or group of adjacent nodes, the memory structure being adapted to allow lookup to identify particular state identifiers, reading of the scores corresponding to the state identifiers, and writing back of the scores to the memory structure after modification of the scores; an accumulator for receiving score updates corresponding to particular state identifiers from a score update generating circuit which generates the score updates using audio input, for receiving scores from the memory structure, and for modifying said scores by adding said score updates to said scores; and a selector circuit for selecting at least one node or group of adjacent nodes of the lexical tree according to said scores.
1. A speech recognition system comprising:
a first programmable device programmed to calculate a feature vector from a digital audio stream, wherein the feature vector comprises a plurality of extracted and/or derived quantities from said digital audio stream during a defined audio time frame;
a second programmable device programmed to calculate distances indicating the similarity between a feature vector and a plurality of acoustic states of an acoustic model wherein said feature vector is received by the second programmable device after it is calculated by the first programmable device; and
a third programmable device programmed to identify spoken words in said digital audio stream using Hidden Markov Models and/or Neural Networks wherein said word identification uses one or more distances that were calculated by the second programmable device, wherein said identification of spoken words uses one or more distances calculated from a first feature vector; and
a search stage for using the calculated distances to identify words within a lexical tree, the lexical tree comprising a model of words.
2. The speech recognition system of claim 1 wherein the time period taken for said identification of spoken words overlaps the time period taken for said second programmable device to calculate one or more distances from a second feature vector.
3. The speech recognition system of claim 2 wherein the first programmable device, the second programmable device, and third programmable device are physically distinct processing circuits.
4. The speech recognition system of claim 3 wherein said physically distinct processing circuits are all formed on a single integrated circuit.
5. The speech recognition system of claim 2 wherein the first programmable device and the second programmable device comprise one physical uni-processor or multi-processor, and the second programmable device is a physically distinct processing circuit.
6. The speech recognition system of claim 5 wherein said one physical uni-processor or multi-processor and said second programmable device are all formed on a single integrated circuit.