Huiqun Deng

Learn More
It is known that convolutional neural networks (CNNs) are efficient for optical character recognition (OCR) and many other visual classification tasks. This paper applies error-correcting output coding (ECOC) to the CNN for segmentation-free OCR such that: 1) the CNN target outputs are designed according to code words of length N; 2) the minimum Hamming(More)
We present a new method for estimating the vocal-tract area functions from speech signals. First, we point out and correct a long-standing sign error in some literature related to the derivation of the acoustic reflection coefficients of the vocal tract from a speech signal. Next, to eliminate the influence of the glottal wave on the estimation of the(More)
This paper shows how to obtain accurate glottal waves via inverse filtering of vowel sounds and how to determine if these glottal waves contain any significant resonance of vocal tracts. We obtain vocal-tract filter (VTF) estimates for the inverse filtering from sustained vowel sounds over closed glottal phases using a new method, which minimizes the(More)
The acoustic transfer functions of external ears with or without headphones affect the features of perceived sounds and vary considerably with listeners and headphones. A method for estimating the frequency responses of external-ear transfer functions from the sound at the entrance of a blocked ear canal (or from the input of a headphone) to the sound at(More)