Method of forming augmented corpus related to articulation disorder, corpus augmenting system, speech recognition platform, and assisting device

A corpus augmenting system, comprising:

• a conversion model;
• a feature acquiring module configured to
• receive a target corpus recorded from a user with articulation disorder and acquire a set of target speech feature data therefrom,
• receive a plurality of training corpora that are related to normal speech, and acquire a plurality of sets of training speech feature data respectively from the training corpora,
• receive an augmenting source corpus that is related to normal speech and acquire a set of augmenting source speech feature data from the augmenting source corpus,
• feed the set of target speech feature data and the sets of training speech feature data into said conversion model for training said conversion model, so as to make said conversion model thus trained capable of converting each of the sets of training speech feature data into a respective set of output data that is similar to the set of target speech feature data, and
• input the set of augmenting source speech feature data into said conversion model thus trained, said conversion model thus trained being configured to convert the set of augmenting source speech feature data into a set of converted speech feature data;
• and
• a waveform reconstruction module configured to synthesize the augmented corpus based on the set of converted speech feature data.