ISCA Archive Interspeech 2015
ISCA Archive Interspeech 2015

Improvements to the pruning behavior of DNN acoustic models

Matthias Paulik

This paper examines two strategies that improve the beam pruning behavior of DNN acoustic models with only a negligible increase in model complexity. By augmenting the boosted MMI loss function used in sequence training with the weighted cross-entropy error, we achieve a real time factor (RTF) reduction of more than 13%. By directly incorporating a transition model into the DNN, which leads to a parameter size increase of less than 0.017%, we achieve a RTF reduction of 16%. Combining both techniques results in a RTF reduction of more than 23%. Both strategies, and their combination, also lead to small but statistically significant word error rate reductions.