Finding Acoustic Regularities in Speech: From Words to Segments

The development of an automatic speech recognizer is typically a highly supervised process involving the specification of phonetic inventories, lexicons, acoustic and language models, along with annot