Back to results
US 8,818,807 B1Granted
Large vocabulary binary speech recognition
OwnerDarrell Poirierinventor
Life of this patent
Estimating the term…
Abstract
This invention describes methods for implementing human speech recognition. The methods described here are of using sub-events that are sounds between spaces (typically a fully spoken word) that is then compared with a library of sub-events. All sub-events are packaged with it's own speech recognition function as individual units. This invention illustrates how this model can be used as a Large Vocabulary Speech Recognition System.
Claims2
A large vocabulary speech recognition system, comprised of;
- a) a stream of audio words in a time order,
- b) a divider, adapted to divide the stream of audio words into individual audio words,
- c) a plurality of binary speech recognition processors, each said processor being adapted to input an individual audio word, and output the text of a single specific assigned word that is assigned to that processor, if and only if the processor recognizes that an audio word presented to it represents that specific assigned word, and if the audio word presented to it is not the assigned word of the processor, the processor does not output the text of the presented word,
- d) a presenter adapted to take each individual audio word from the divider, and presents it simultaneously to each of the plurality of binary speech recognition processors for input and processing by that binary speech recognition processor,
- e) an assembler adapted to gather the text words from binary speech recognition processors and assemble the text words into a second stream of text words, in the same time order as the first stream of audio words.
2 more elements
A method for achieving large vocabulary speech recognition, comprised the steps of;
- a) inputing a first stream of audio words in a time order,
- b) dividing the stream of audio words into individual audio words,
- c) presenting each audio word simultaneously to a plurality of binary speech recognition processors for processing, each said processor being adapted to input an individual audio word, and output the text of a single specific assigned word that is assigned to that processor, if and only if the processor recognizes that the audio word presented to it represents that specific assigned word, and if the audio word presented to it is not the assigned word of the processor, the processor does not output the text of the presented word,
- d) gathering the text words from binary speech recognition processors and assembling the text words into a stream of text words, in the same time order as the stream audio words.
1 more element
Dependencies are inferred from the claim text.