Abstract: Systems and methods for machine-learning based multi-lingual pronunciation generation are disclosed. A method for machine-learning based multi-lingual pronunciation generation may include: (1) training a language origin prediction machine learning model; (2) training a pronunciation generator machine learning model; (3) receiving, by a pronunciation computer program, a word for pronunciation guidance; (4) predicting, by the pronunciation computer program and using the trained language origin prediction machine learning model, a language origin of the word; (5) predicting, by the pronunciation computer program and using the trained pronunciation generator machine learning model and the language origin, a syllable-by-syllable pronunciation for the word; and (6) returning, by the pronunciation computer program, the syllable-by-syllable pronunciation.

Abstract: Systems and methods for efficient speech representation are disclosed. In one embodiment, a method for efficient speech representation may include training a teacher model using training data to get embeddings from intermediate and/or final layers of the teacher model; training a student model using training data processed with audio distortion to have outputs matching the embeddings from the teacher model; and injecting known hand-crafted audio features into intermediate or final layers of the student model.

Abstract: Systems and methods for simultaneous network pruning and parameter optimization are disclosed. A method may include: (1) receiving a network to optimize, the network comprising a plurality of layers; (2) selecting layer pruning and/or channel pruning for the layers within the network; (3) providing a gating module at layers within the network, wherein each gating module opens or closes a gate in the gating module based on an output of a binary head; (4) training parameters for the network; (5) extracting gate open/close features from the gating modules; (6) optimizing a loss function for the network using a polarization regularizer to reach a consensus static sub-network; and (7) updating parameters for each gating module in the network consistent with the consensus static sub-network.