www.sciedu.ca/air Artificial Intelligence Research 2015, Vol. 4, No. 2 ORIGINAL RESEARCH Cross-language phoneme mapping for phonetic search keyword spotting in continuous speech of under-resourced languages Ella Tetariy 1 , Yossi Bar-Yosef 2 , Vered Silber-Varod 1 , Michal Gishri *1 , Ruthi Alon-Lavi 2 , Vered Aharonson 1 , Irit Opher 2 , Ami Moyal 1 1 Afeka Academic College of Engineering, Afeka Center for Language Processing, Tel Aviv, Israel 2 NICE Systems Ltd., Ra’anana, Israel Received: March 29, 2015 Accepted: May 19, 2015 Online Published: June 25, 2015 DOI: 10.5430/air.v4n2p72 URL: http://dx.doi.org/10.5430/air.v4n2p72 Abstract As automatic speech recognition-based applications become increasingly common in a wide variety of market segments, there is a growing need to support more languages. However, for many languages, the language resources needed to train speech recognition engines are either limited or completely non-existent, and the process of acquiring or constructing new language resources is both long and costly. This paper suggests a methodology that enables Phonetic Search Keyword Spotting to be implemented in a large speech database of any given under-resourced language using cross-language phoneme mappings to another language. The phoneme mapping enables a speech recognition engine from a sufficiently resourced and well-trained source language to be used for phoneme recognition in the new target language. The keyword search is then performed over a lattice of target language phonemes. Three cross-language phoneme mapping techniques are examined: knowledge-based, data-driven and phoneme recognition performance-based. The results suggest that Phonetic Search Keyword Spotting based on the cross-language phoneme mapping approach proposed herein can serve as a quick initial solution for validating keyword spotting applications in new, under-resourced languages. Key Words: Cross-language phoneme mapping, Keyword spotting, Spoken term detection, Phonetic search 1 Introduction Speech indexing and retrieval tools have become increas- ingly crucial in coping with the constant accumulation of massive amounts of digital audio and video data. In par- ticular, speech recognition technology is frequently used in Keyword Spotting (KWS)-based applications to enable spe- cific words to be identified out of a stream of continuous. [1] KWS-based applications, in turn, are often used by call cen- ters and security-intelligence organizations for categorizing calls or searching speech databases, or by companies of- fering multi-media search applications on the internet or in enterprise markets. Such applications can be developed quickly for languages with sufficient available Language Resources (LRs). Supporting an under-resourced language, however, generally requires a long and costly preliminary process of collecting speech and text databases in order to train acoustic and language models, in addition to compil- ing a large vocabulary pronunciation lexicon. Yet, in spite of these challenges, there seems to be a growing demand for providing rapid support for under-resourced languages, * Correspondence: Michal Gishri; Email: michalg@afeka.ac.il; Address: Afeka Academic College of Engineering, Afeka Center for Language Processing, 38Mivtsa Kadesh St. Tel-Aviv, 6998812, Israel. 72 ISSN 1927-6974 E-ISSN 1927-6982