Method for automatic identification of isolated words based on image analysis time characteristics

摘要

method for automatic identification of isolated words based on the analysis of characteristics of time is suchin the initial processing block is fixed level zero signal output transducer collected from microphone output transducer then local minima are located and fixed the boundaries of individual periods of fundamental tone krtaniowego and recorded to the samples b u0119du0105cych borders fundamental periods.what is carried out imposing and coding networks and the results obtained for the matrix, in the next stage are designated the similarity and compatibility on the basis of the comparison matrix with the number of each of the preceding and following the matrix find u0105cymi in memory, the number of similar matrix is remembered as factor conformitythe next action is to find fonemu00f3w szumowych, remembering their location on a time axis and the exclusion of further analysis and automatic location of the first fonemu containing basic tone krtaniowego periods after which time zones are traced, transitional and another fo nemy sound after the obtained structure fonemowa is compared with base modelsthe analysis is also subjected to envelope the expressions and compared with the base patterns enveloping final step is explicit identification result after finding structure in accordance with one of the orders of fonemowej database commands and expressions odpowiad envelope patterns aju0105cej specimen envelope for that instruction.