The transmission of spectral information consumes a large portion of total bit rate in medium-to-low bit rate speech coding. The conventional coding methods of LSP parameters generate redundant spectral information or spectral distortion by the fixed update rate of LSP parameters independent of the order of coefficients and phonetic context. We propose a multiple type frame segmentation (MTFS) method which allows various types of two-dimensional segmentation of speech frames to save the transmission rate of the LSP parameters without increasing the spectral distortion. The intra-frame spectral distortion (IFSD) is defined to measure the spectral distortion of the reconstructed spectrum. The proposed method generates a less distorted spectrum with fewer bits compared with the conventional single type frame segmentation (STFS) method.

implement Korean This paper describes a way of neural hardware implementation with the analog-digital mixed mode neural chip. The full custom neural VLSI of Universally Reconstructible Artificial Neural network (URAN) is used system. A to multi-layer perceptron with is trained successfully under the limited accuracy in computations. The network with a large frame input layer is tested to recognize spoken korean words at a forward retrieval. Multichip hardware module is suggested with eight chips or more for the extended performance and capacity.