Automatic Speech Attribute Transcription (ASAT), an ITR project sponsored under the NSF grant (IIS-04-27113), and Spoken Information Retrieval by Knowledge Utilization in Statistical Speech Processing (SIRKUS), a project funded by the VERDIKT programme at the Research Council of Norway, are two research projects carried out at Georgia Institute of Technology and at Norwegian University of Science and Technology, respectively, with the purpose of investigating and developing new paradigms for speech recognition that have the capability of bridging the gap between machine and human performance. These projects approach speech recognition from a more linguistic perspective: unlike traditional ASR systems, humans detect acoustic and auditory cues, weigh and combine them to form theories, and then process these cognitive hypotheses until linguistically and pragmatically consistent speech understanding is achieved. A major goal of the ASAT/SIRKUS paradigms is to develop a detection-based approach to automatic speech recognition (ASR) based on attribute detection and knowledge integration. We report on progress of these two projects on two different tasks, namely the cross-language and language universal attribute/phone recognition task, and the language identification (LID) task.

A survey on recent progress in the ASAT/SIRKUS paradigm

SINISCALCHI, SABATO MARCO;
2010

Abstract

Automatic Speech Attribute Transcription (ASAT), an ITR project sponsored under the NSF grant (IIS-04-27113), and Spoken Information Retrieval by Knowledge Utilization in Statistical Speech Processing (SIRKUS), a project funded by the VERDIKT programme at the Research Council of Norway, are two research projects carried out at Georgia Institute of Technology and at Norwegian University of Science and Technology, respectively, with the purpose of investigating and developing new paradigms for speech recognition that have the capability of bridging the gap between machine and human performance. These projects approach speech recognition from a more linguistic perspective: unlike traditional ASR systems, humans detect acoustic and auditory cues, weigh and combine them to form theories, and then process these cognitive hypotheses until linguistically and pragmatically consistent speech understanding is achieved. A major goal of the ASAT/SIRKUS paradigms is to develop a detection-based approach to automatic speech recognition (ASR) based on attribute detection and knowledge integration. We report on progress of these two projects on two different tasks, namely the cross-language and language universal attribute/phone recognition task, and the language identification (LID) task.
9781424462445
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/11387/77332
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? ND
social impact