A comprehensive study and benchmark of 4 different Prediction Networks in RNN-T for offline
Two methods allow almost full recovery of the performance loss due to data deidentification
Enabling a more natural doctor-patient interaction with Ambient Intelligence.
The ChannelAugment technique improves robustness of multi-channel models
Dual-encoder architecture enables more accurate and natural speech interfaces
Nuance’s deep learning technology nearly halves the baseline error rate on named entities
Deep learning can be used for more accurate hands-free speech recognition
We present a framework for estimating acoustic properties of a signal using deep learning
New semi-supervised learning algorithm yields similar accuracy with 75% less labeled data
Discover how speech recognizers can operate in conditions unseen in training data.
End-to-end learning can adapt the way computers perceive various aspects of speech.
