Skip to main content
Go to the home page of the European Commission (opens in new window)
English en
CORDIS - EU research results
CORDIS
Content archived on 2024-04-16

Multilingual Speech Input/Output: Assessment, Methodology and Standardisation

Objective

The objective of SAM was to develop methodologies, tools and databases for the assessment of speech synthesis and recognition systems in applications where multilingual performance is required from the same basic equipment.
The basic aims are to define, establish and promulgate standards in respect of speech technology assessment. Evaluation and understanding of the performance of speech input/output systems are linked with these aims, and are therefore important tools.

The objective of the project is to develop methodologies, tools and databases for the assessment of speech synthesis and recognition systems in applications where multilingual performance is required from the same basic equipment.

The activities in the main phase of the project focus around 3 major areas:
speech input assessment (including the setting up of tests, their scoring, and the delineation of performance factors involving recognizer tests that have been defined and run on the project's standard personal computer (PC) based workstations, SESAM, using standardized scoring software and cross language work is also being directed towards the definition of recognizer sensitive sensitivity analysis);
speech output assessment (both objective and subjective, including standard multilanguage tests for evaluation at the segment and suprasegmental level have been developed to run on the workstation, using automatic response scoring procedures, together with overall quality measures);
enabling technologies (the specification of speech databases across all the project languages, the definition of uniform protocols for collection and management of speech databases in a standard format and the introduction of agreed methods for multilingual annotation are all examples of work in this area).

The European SESAM workstation has been provided with tools for speech labelling, analysis and input/output assessment, and is already in use. Basic speech knowledge at the physical, phonetic and phonological levels is organized for the language of the project.
The consortium included participants from six EC and two EFTA member states, so that techniques for assessing speech synthesisers and recognisers for eight languages would be produced. The wide participation basis also gives a strong support to the final recommendations. Furthermore, close ties have been established with related national projects in the participating countries, many of which are now using SAM standards.

Fields of science (EuroSciVoc)

Programme(s)

Topic(s)

Data not available

Call for proposal

Data not available

Funding Scheme

Data not available

Coordinator

Birkbeck College, University of London
EU contribution
No data
Address
Malet Street, Bloomsbury
WC1E 7HX London
United Kingdom
Total cost
No data

Participants (15)