| 摘要 |
<p>PURPOSE:To practically improve the recognition performance of an acoustic recognizing part by using a reverse propagation network model to correct a time series of symbols including erroneous recognition obtained as the acoustic recognition result. CONSTITUTION:A symbol string as the acoustic recognition result is stored in an input buffer part 1, and an input window part 2 successively segments a fixed-length symbol string while shifting the start point by every one symbol and inputs it to a reverse propagation network model part 3. A rewrite part 5 rewrites a corresponding symbol in the input buffer part 4 with the output symbol each time when the model part 3 outputs the inference result of the input. The output of the reverse propagation network model part 3 is stored in an output buffer part 4, and an input control part 5 shifts the start position of the input window part 2 by one symbol to perform the next correcting operation each time when the reverse propagation network model part 3 outputs one symbol and it is written back in the input buffer part 1. When detecting that correction is performed up to the last symbol in the input buffer part 1, a correction result output part 6 outputs stored contents of the output buffer part 4. Thus, the recognition performance of the acoustic recognizing part is improved.</p> |
