Deliverables Methods and rules for positioning engineering design objects by speech commands Formulas and methods were developed for calculating positions of geometrical objects from natural language specification (e.g., above, on right, etc.). There is a potential for incorporation into engineering design systems, educational systems, or publishing systems. The methods have been implemented and tested for rectangular, cylindrical and bent cylindrical parts in 2D and 3D demonstrators. Expected benefits: - Ease of expressing position relationship of one part relative to another by using natural language instead of coordinates. - More precise positioning possible than with pointing. Procedure for �speech objectification� of engineering design objects These are methods for choosing attributes of geometrical objects that can be calculated from natural language specifications. There is a potential for incorporation into engineering design systems, educational systems, or publishing systems. The procedure has been implemented and tested for rectangular, cylindrical and bent cylindrical parts in 2D and 3D demonstrators. Expected benefits: Ease of expressing object rules and constraints using natural language. Lexical entries for defining and positioning engineering design entities Lexica is developed for speech commands for engineering design tasks. Lexica can be incorporated into engineering design systems. It is currently incorporated into the demonstrator system. Expected benefits: Ease of expressing object constraints using natural language. Report on results of dysarthria test and acoustic analysis for 12 dysarthric speakers The report describes some aspects of dysarthric speech by means of acoustic analysis. Parameters such as pitch and syllable repetition rate are included in the analysis. Parameters that are measured which are of potential importance to the recogniser are speaking rate, frequency of pauses (dysfluencies) and articulation. Deviations from the normal range and mean have been detected in all parameters. No significant correlation between the parameters has been found concerning fluency vs. intelligibility and articulation deviation score but all symptoms occur with different degree and quality in all speakers. This finding indicates the complexities in predicting which individuals of the dysarthric population will be intelligible (including understandable by ASR systems) and which are not. One of the critical questions that arise is whether the parameters that are essential to human intelligibility or correctly labeled by acoustic/phonetic methods also yield the optimal machine recognition result. Broker for speech user interface The very simple architecture of the Broker software provides a lightweight, simple-to-use system. The communication uses only text (no binary data) to achieve platform independence and to facilitate debugging. The servers can be restarted automatically when needed, or if they had crashed. A central server, named the Broker, is used; the server is similar to the Open Agent Architecture blackboard. The Broker routes calls between modules. It has knowledge of which servers exist, and how to start them. All other programs connect to this server using the Transaction Control Protocol (TCP). The use of text communication and TCP has made it very easy to implement Broker components in the scripting language. It is possible to connect to the Broker with Telnet and manually to interact with the Broker modules. Predictive scale for assessing degree of 'risk' for individuals and vocations of known voice load A vocal risk scale was compiled to assess the level of those individuals who are at-risk for vocal health. The scale is a self-assessment scale which consists of 25 yes/no questions related to levels of voice use (vocal load), lifestyle, work and vocal health. A point is scored for every affirmative ('yes') answer to give a vocal risk score. Higher vocal risk scores are indicative of those who are at most risk for vocal health. The vocal risk scale has been tested on 26 subjects, and has shown to be accurate in assessing the vocal loads of most of these subjects. The vocal risk scale was used to compile vocal risk scores for the 26 subjects. These scores assisted in the categorisation of subjects into heavy vocal load (at-risk) and normal vocal load groups for the assessment of vocal behaviour of subjects using discrete and continuous voice input software. A list of recommendations was subsequently suggested to improve speech recognition levels and hence vocal care. Searching for OpenAIRE data... There was an error trying to search data from OpenAIRE No results available
