Littera Deusto

Modern Languages, Basque Studies and Humanities

Research on Human Language Technologies

mayo 22nd, 2008 · No hay Comentarios

As requested in Questionnaire 2 (Human Language Technology Course), I have selected several research topics that are being currently developed and investigated in the Human Language Technologies centres and laboratories listed in the previous article. The Pattern Recognitionand Human Language Technology (PRHLT) research group, for example, is currently working on Machine Translation, wich is described in Wikipedia as a sub-field of computational linguistics and is sometimes referred to by the abbreviation MT.  MT investigates the use of computer software to translate text or speech from one natural language to another. At its basic level, MT performs simple substitution of words in one natural language for words in another. Using corpus techniques, more complex translations may be attempted, allowing for better handling of differences in linguistic typology, phrase recognition, and translation of idioms, as well as the isolation of anomalies.In PRHLT they work on machine translation through speech-to-speech translation or text-to-text translation for limited domains. The basis of their machine translation systems are finite-state and statistical transducers.As can be seen in wikipedia, a finite state transducer (FST) is a finite state machine with two tapes: an input tape and an output tape.These models can be learnt automatically from real examples of translation. The possible applications for this technology are translation of technical reports, hotel services, etc.

In the National Centre for Language Technology , Machine Translation and Translation Technology is also among the several topics that constitute an object of research. In this centre they focus on translation tools such as Statistical and Rule-Based MT (SMT, RBMT), Example-Based MT (EBMT), Translation Memories (TMs), Boosting Existing MT Systems, Machine-Aided Translation (MAT) and Computer-Aided Translation (CAT). In the articles related to questionnaire three I will concentrate on this topic in a more specific manner.

Another subject of research on Human language Technologies is Speech Technology, wich encompasses many subdisciplines and encorporates much technology from other areas.The National Centre for Language Technology, for example, is studing subdisciplines such as Speaker Characterisation, Audio Classification, Retrieval and Coding, and Human Computer Interfaces (HCIs). This topics are directly related to Speech Technology, that has several applications such as automatic speech recognition (ASR) wich means in simple terms, talking to machines; and speech generation (synthesis) or getting machines to talk. As it says in the website of this centre, Speech Analysis is particularly important in helping us better understand the production of human speech which in turn helps us improve Speech Technology in general, so this also constitutes an important part of this kind of research.

Automatic Speech recognition (ASR) is related to Speech Technology, as mentioned in the preceding paragraph. This technology allows a computer to identify the words that a person speaks into a microphone or telephone, or as it says in Wikipedia, The Free Encyclopedia: “ASR converts spoken words to machine-readable input”.This topic appears on the list of current research projects that are being developed by the Pattern Recognition and Human Language Technology (PRHLT) research group. Under the title of Automatic Speech Recognition and Understanding, the project consist in decoding the speech utterances into strings of words or into strings of semantic units, using finite-state grammars as the basis of such systems. These finite-state grammars are learnt automatically from real examples of utterances or text. There are many applications for this technology, including voice dialing, call routing, domotic appliance control and content-based spoken audio search, simple data entry, preparation of structured documents, speech-to-text processing (word processors or emails), and in aircraft cockpits (Direct Voice Input).        ; The PRHLT Group, as can be seen on their website, focuses on some applications such as telephone exchange services, device control by voice, information queries, etc.







  • Etiquetas