Punctuation Prediction Model for Conversational Speech
This work addresses the lack of punctuation in ASR outputs, which can confuse human readers and NLP algorithms, but it is incremental as it applies existing neural network methods to a specific dataset.
The authors tackled the problem of automatic punctuation prediction for conversational speech, which is missing from typical ASR systems, by training CNN and BLSTM models on the Fisher corpus with time-aligned transcripts and pre-trained embeddings. They found that CNNs achieved better precision, especially for question marks, while BLSTMs had better recall and fewer overall errors.
An ASR system usually does not predict any punctuation or capitalization. Lack of punctuation causes problems in result presentation and confuses both the human reader andoff-the-shelf natural language processing algorithms. To overcome these limitations, we train two variants of Deep Neural Network (DNN) sequence labelling models - a Bidirectional Long Short-Term Memory (BLSTM) and a Convolutional Neural Network (CNN), to predict the punctuation. The models are trained on the Fisher corpus which includes punctuation annotation. In our experiments, we combine time-aligned and punctuated Fisher corpus transcripts using a sequence alignment algorithm. The neural networks are trained on Common Web Crawl GloVe embedding of the words in Fisher transcripts aligned with conversation side indicators and word time infomation. The CNNs yield a better precision and BLSTMs tend to have better recall. While BLSTMs make fewer mistakes overall, the punctuation predicted by the CNN is more accurate - especially in the case of question marks. Our results constitute significant evidence that the distribution of words in time, as well as pre-trained embeddings, can be useful in the punctuation prediction task.