Working with Visionaries on the Frontlines of Scientific Progress Worldwide
Nuance Foundation Grants

Karlsruhe Institute of Technology (KIT): Rapid Language Adaptation - An investigation on multilingual Bottle-Neck feature and its application on new languages

The project aims at applying Multi-Layer Perceptrons (MLPs, aka Deep Belief Networks, DBNs) at multi-lingual ASR. The approach tries to re-use features learned on one or more other languages when training as ASR system for a new language, an approach that has been shown to be successful before.

Read More

DFKI: Exploiting User Turn Segmentation in Polylogue

Along conversations we are exposed to various external factors that might have strong influence on our conversational act in a specific moment. In terms of the behavior how natural language is used, this is all the more true if we look at the phenomena of words re-usage in conversations.

Read More

DFKI: Multi-modal interaction with distant objects using eye gaze and multi-touch input

Tabletop interaction with objects in and out of reach is a common real world as well as virtual task. Gaze can be used as additional input modality in order to support search, selection and manipulation of distant objects on digital tabletop.

Read More

University of Cambridge: Confidence Estimation using Hidden-state CRF Models with Word and Phone-level Features

Speech recognition technology allows us to interact with machines in a natural way by speaking to them. This interaction may be for the purposes of dictation, accessing and filtering content available on the internet, controlling a user interface or a number of other applications.

Read More

Stanford University: Dialog Access to Sentiment-Aware Product and Restaurant Information on Mobile Platforms

The project tries to improve mobile access to information over the state of the art, which is typically using a speech interface to just access a general search engine (like Google or Yelp). Instead it is proposed here to take sentiment analysis into the mix and also analyze user provided texts that are available on e.g. restaurant portals to provide more meaningful feedback to the users.

Read More

University of Toronto: A recognition-by-synthesis architecture for dysarthric speech using deep-belief networks

Approximately 10% of North Americans have some sort of communication disorder. It is imperative that technology is used to mitigate difficulties these individuals have in being understood. Unfortunately, speech recognition designed for individuals with disabilities is relatively rare and the utility of the technology remains prohibitively poor for this population.

Read More

McGill University: Efficient Manifold-Constrained Discriminative Acoustic Modeling for Automatic Speech Recognition

The proposed project will investigate techniques that facilitate more efficient use of speech data for acoustic model training in ASR. The first project goal is to enable efficient configuration of initial ASR systems in previously unseen languages. This will be done by employing techniques for leveraging speech data from multiple languages in order to configure an initial ASR system in a target under-resourced language.

Read More

Nuance Foundation News

No recent news at this time.