Research on Chinese Digit Speech Recognition Based on Multi-Weighted Neural Network

This paper proposes a new approach for speaker-independent continuous Chinese digit speech recognition based on multi-weighted neural network. By analyzing the geometrical meaning of the multi-weighted neuron function and on the principle of continuity point of view, this approach constructs a multi-weighted neural network. This network shows better recognition results when applied in digit speech recognition. Its correct recognition rate is higher than HMM.