Towards Using Context-Dependent Symbols in CTC Without State-Tying Decision Trees
This work addresses a bottleneck in neural speech recognition systems by enabling direct training with context-dependent symbols, making it incremental but impactful for the field.
The paper tackles the problem of training CTC networks with context-dependent outputs by addressing overfitting from frame-level normalization and poor generalization with numerous lexical units, achieving parameter savings and removing the need for bootstrapping from GMM-HMM systems.
Deep neural acoustic models benefit from context-dependent (CD) modeling of output symbols. We consider direct training of CTC networks with CD outputs, and identify two issues. The first one is frame-level normalization of probabilities in CTC, which induces strong language modeling behavior that leads to overfitting and interference with external language models. The second one is poor generalization in the presence of numerous lexical units like triphones or tri-chars. We mitigate the former with utterance-level normalization of probabilities. The latter typically requires reducing the CD symbol inventory with state-tying decision trees, which have to be transferred from classical GMM-HMM systems. We replace the trees with a CD symbol embedding network, which saves parameters and ensures generalization to unseen and undersampled CD symbols. The embedding network is trained together with the rest of the acoustic model and removes one of the last cases in which neural systems have to be bootstrapped from GMM-HMM ones.