Low-resource spoken keyword search strategies in georgian inspired by distinctive feature theory

Nancy F. Chen, Boon Pang Lim, Van Hai Do, Van Tung Pham, Chongjia Ni, Haihua Xu, Mark Hasegawajohnson, Wenda Chen, Xiong Xiao, Sunil Sivadas, Eng Siong Chng, Bin Ma, Haizhou Li

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

We present low-resource spoken keyword search (KWS) strategies guided by distinctive feature theory in linguistics to conduct data selection, feature selection, and transcription augmentation. These strategies were employed in the context of the 2016 NIST Open Keyword Search Evaluation (OpenKWS16) using conversational Georgian from the IARPA Babel program. In particular, we elaborate on the following: (1) We exploit glottal-source-related acoustic features that characterize Georgian ejective phonemes ([+constricted glottis], [+raised larynx ejective] specified in distinctive feature theory). These features complement standard acoustic features, leading to a relative fusion gain of 11.9%. (2) We use noisy channel models to incorporate probabilistic phonetic transcriptions from mismatched crowdsourcing to conduct transfer learning to improve KWS for extremely under-resourced conditions (24 min of transcribed Georgian), achieving a relative improvement of 118% over the baseline and a relative fusion gain of 32%.(3) Using distinctive feature analysis, we select a compact subset of source languages used in past evaluations to ensure high phonetic coverage for cross-lingual acoustic modeling when only limited system development time and computational resources are available. This strategy leads to comparable performance to using all available linguistic resources when only 1/3 of the source languages were chosen.

Original languageEnglish (US)
Title of host publicationProceedings - 9th Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2017
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1322-1327
Number of pages6
ISBN (Electronic)9781538615423
DOIs
StatePublished - Feb 5 2018
Externally publishedYes
Event9th Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2017 - Kuala Lumpur, Malaysia
Duration: Dec 12 2017Dec 15 2017

Publication series

NameProceedings - 9th Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2017
Volume2018-February

Other

Other9th Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2017
CountryMalaysia
CityKuala Lumpur
Period12/12/1712/15/17

    Fingerprint

ASJC Scopus subject areas

  • Artificial Intelligence
  • Human-Computer Interaction
  • Information Systems
  • Signal Processing

Cite this

Chen, N. F., Lim, B. P., Do, V. H., Pham, V. T., Ni, C., Xu, H., Hasegawajohnson, M., Chen, W., Xiao, X., Sivadas, S., Chng, E. S., Ma, B., & Li, H. (2018). Low-resource spoken keyword search strategies in georgian inspired by distinctive feature theory. In Proceedings - 9th Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2017 (pp. 1322-1327). (Proceedings - 9th Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2017; Vol. 2018-February). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/APSIPA.2017.8282237