Neural network vocoder training method based on short-time spectrum consistency

Abstract

The invention discloses a neural network vocoder training method based on short-time spectrum consistency. The method comprises steps of training an amplitude spectrum predictor through employing natural acoustic features and a natural logarithm amplitude spectrum, and training a phase spectrum predictor through employing the natural logarithm amplitude spectrum, a natural fundamental frequency and a natural waveform; connecting the trained amplitude spectrum predictor and phase spectrum predictor, and training the connected amplitude spectrum predictor and phase spectrum predictor through the natural acoustic features, the natural logarithm amplitude spectrum, the natural fundamental frequency and the natural waveform. According to the neural network vocoder training method provided by the invention, firstly, the amplitude spectrum predictor and the phase spectrum predictor are respectively trained, and finally, the short-time spectrum consistency loss function is added to jointly train the amplitude spectrum predictor and the phase spectrum predictor, so a problem of inconsistency of short-time spectrums formed by predicted amplitude spectrums and phase spectrums can be greatly improved; and the quality of the synthesized voice is improved.

Description

technical field [0001] The invention relates to the technical field of speech signal processing, in particular to a neural network vocoder training method and a speech synthesis method based on short-time spectrum consistency. Background technique [0002] Speech synthesis, which aims to enable machines to speak fluently and naturally like humans, has benefited many voice-interactive applications, such as intelligent personal assistants and robots. Currently, statistical parametric speech synthesis (SPSS) is one of the mainstream methods. [0003] Statistical parametric speech synthesis utilizes an acoustic model to model the relationship between text features and acoustic features, and utilizes a vocoder to obtain speech waveforms given predicted acoustic features. The performance of the vocoder can significantly affect the quality of synthesized speech. Traditional vocoders such as STRAIGHT and WORLD are widely used in current SPSS systems. However, these traditional vo...

AI Technical Summary

Problems solved by technology

[0005] On the basis of the original training method of the neural network vocoder, the present invention designs a short-term spectrum consistency loss function to jointly train the amplitude spectrum predictor and the phase spectrum predictor in the neural network vocoder, so as to reduce the predicted amplitude spectrum The problem of inconsistency in the short-term spectrum combined with the phase spectrum improves the quality of synthesized speech

Images