Manual Cued Speech (MCS) is an effective method of communication by the deaf and hearing-impaired. We first describe our work on assessing the feasibility of automatic determination and presentation of cues without intervention by the speaker. The conclusions of this study are then applied to the design and implementation of a prototype automatic cueing system using HMM-based automatic speech r...

This study investigates the process of generalizing a learned sub-lexical contrast across speakers of different non-native accents of English. We examine the generalization of a novel cue (voicing-cued release) that is non-contrastive in English, but contrastive in the manipulated speech of our L2 speakers of English, to a speaker with the same or different L1 as our training speakers (Exp. 1)....

In reading research, a longstanding question is whether any stages of lexical processing require central attention, and whether such potential demands are frequency-sensitive. In the present study, we examined the allocation of cognitive effort in lexical processing by examining pupil dilations and naming latencies in a modified delayed naming procedure. In this dual-task/change procedure, part...

Discourse cues (e.g., because, since, therefore, and so) express structural and semantic relationships between parts of an explanation, and are used extensively by human tutors. Previous research by Moser and Moore (in preparation) revealed patterns of cue occurrence, placement, and selection in the speech of expert tutors during tutorial dialogues. An experiment measuring both reading times an...

We present a novel method for generating additional pseudo-articulator trajectories suitable for use within the framework of a stochastically trained speech production system recently developed at CUED. The system is initialised by inverting a codebook of (articulator, spectral vector) pairs, and the target positions for a set of pseudo-articulators and the mapping from these to speech spectral...

In recent years, recurrent neural network language models (RNNLMs) have become increasingly popular for a range of applications including speech recognition. However, the training of RNNLMs is computationally expensive, which limits the quantity of data, and size of network, that can be used. In order to fully exploit the power of RNNLMs, efficient training implementations are required. This pa...

A cross-modal dual attention experiment was completed by 198 undergraduates in three blocks that each consisted of an orientation task and a concurrent listening task. For the orientation task, participants located regions on an LCD that were cued by speech or one of four types of symbolic auditory cues (i.e. earcons); the concurrent task required participants to listen to and answer questions ...

Two word-spotting experiments are reported that examine whether the Possible-Word Constraint (PWC) is a language-speciŽc or languageuniversal strategy for the segmentation of continuous speech. The PWC disfavours parses which leave an impossible residue between the end of a candidate word and any likely location of a word boundary, as cued in the speech signal. The experiments examined cases wh...

We present a global overview of imageand video-processing-based methods to help the communication of hearing impaired people. Two directions of communication have to be considered: from a hearing person to a hearing impaired person and vice versa. In this paper, firstly, we describe sign language (SL) and the cued speech (CS) language which are two different languages used by the deaf community...