The present invention provides a method and an apparatus for generating a phoneme conversion model based on artificial intelligence. In the training method, a neural network is applied to perform phoneme conversion for each word in training data, Performing random screening on hidden node nodes of the neural network in proportion to a preset node to obtain a retaining node for training each word; Training the respective word by applying a sub-neural network corresponding to the retaining node, and updating a related weight of each retaining node of the sub-neural network; And finally performing a mean value process on the related weight of each retaining node of all the sub-neural networks to generate a phoneme conversion model. Through the method and apparatus for generating a phoneme conversion model based on the artificial intelligence provided by the present application, the learning ability and the generalization ability of the phoneme conversion model are improved to avoid over-sum phenomenon.
展开▼