Parallelism, hierarchy, scaling in time-delay neural networks for spotting Japanese phonemes CV-syllables

To extend the performance of TDNNs (time-delay neural networks) to all phoneme recognition and word/continuous speech recognition, the authors present several techniques. First, they show that it is possible to scale up the TDNN to a large phonemic TDNN aimed at discriminating all phonemes without loss of recognition performance and without excessive training tokens. Second, the authors propose fast backpropagation learning methods which make it possible to train a large phonemic TDNN within 1.5 hours. Finally, they show several methods for spotting Japanese CV syllables/phonemes in input speech based on TDNNs: they constructed a TDNN which can discriminate a single CV syllable or phoneme. Syllable and phoneme spotting experiments show excellent results, with syllable and phoneme spotting rates of better than 96.7% and 92% correct, respectively.<>