On using units trained on foreign data for improved multiple accent speech recognition

Bartkova, Katarina; Jouvet, Denis

doi:10.1016/j.specom.2006.12.009

Cited by 15 publications

(16 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…(Bonaventura et al, 1998;Wester et al, 2000;Schaden, 2003a,b;Bartkova and Jouvet, 2007), or they can be learned automatically from correspondences between base forms and observed pronunciations. This so-called data-driven approach is exemplified by e.g.…”

Section: Lexical Modeling For Lvcsrmentioning

confidence: 99%

Improving proper name recognition by means of automatically learned pronunciation variants

Réveil¹,

Martens²,

Heuvel³

2012

Speech Communication

View full text Add to dashboard Cite

This paper introduces a novel lexical modeling approach that aims to improve large vocabulary proper name recognition for native and non-native speakers. The method uses one or more so-called phoneme-to-phoneme (P2P) converters to add useful pronunciation variants to a baseline lexicon. Each P2P converter is a stochastic automaton that applies context-dependent transformation rules to a baseline transcription that is generated by a standard graphemeto-phoneme (G2P) converter. The paper focuses on the inclusion of different types of features to describe the rule context -ranging from the identities of neighboring phonemes to morphological and even semantic features such as the language of origin of the name -and on the development and assessment of methods that can cope with crosslingual issues. Another aim is to ensure that the proposed solutions are applicable to new names (not seen during system development) and useful in the hands of product developers with good knowledge of their application domain but little expertise in automatic speech recognition (ASR) and speech corpus acquisition. The proposed method was evaluated on person name and geographical name recognition, two economically interesting domains in which nonnative speakers as well as non-native names occur very frequently. For the recognition experiments a state-of-the-art commercial ASR engine was employed. The experimental results demonstrate that significant improvements of the recognition accuracy can be achieved: large gains (up to 40% relative) in case prior knowledge of the speaker tongue and the name origin is available, and still significant gains in case no such prior information is available.

show abstract

Section: Lexical Modeling For Lvcsrmentioning

confidence: 99%

Improving proper name recognition by means of automatically learned pronunciation variants

Réveil¹,

Martens²,

Heuvel³

2012

Speech Communication

View full text Add to dashboard Cite

show abstract

“…In [15,16], the more challenging case of multiple foreign accents was considered. French commands and expressions uttered by speakers from 24 different countries were recognised using a baseline French system, and a multilingual system that was obtained by supplementing the French acoustic models with three foreign (English, German and Spanish) acoustic model sets that were trained on speech from the corresponding languages.…”

Section: Acoustic Modeling Approachesmentioning

confidence: 99%

Lexical Modeling for Proper name Recognition in Autonomata Too

Réveil

Martens

Heuvel

et al. 2012

Essential Speech and Language Technology for Dutch

View full text Add to dashboard Cite

Points of Interest business applications are strongly emerging on the ICT market, in particular in high-end navigation systems. For cars for instance, there is a high safety issue, and voice-driven navigation systems are appealing because they offer hands-and (partly) eye-free operation. However, Points of Interest like company names, hotel and restaurant names, names of attraction parks and museums, etc., often contain non-native parts. Moreover, the targeted application must be usable by non-native as well as native speakers. This means that there are considerable crosslingual effects to cope with, which implies that the challenges for the automatic recogniser are high. At the start of the project (February, 2008) there was indeed substantial evidence [1-8] that state-of-the-art ASR technology was not yet good enough to enable a sufficiently reliable voice-driven POI business service.The general project aim was therefore to improve name recognition accuracy by better coping with the large degree of variations observed in the POI pronunciations. The specific aim was to improve the recognition of (1) native Dutch/Flemish pronunciations of Dutch/Flemish POI, (2) native Dutch/Flemish pronunciations of foreign POI, and (3) non-native pronunciations of Dutch and Flemish POI. An important constraint was that the envisaged approach would have to be easily transferable from one application domain (e.g. car navigation) to another (e.g. B. Réveil ( ) J.-P. Martens Ghent University, ELIS-DSSP, Sint-Pietersnieuwstraat 41,

show abstract

“…These rewriting rules include phonetic and graphemic contexts. A recent work presented in Bartkova and Jouvet (2006) and Bartkova and Jouvet (2007) targets multilingual non-native speech recognition through modeling based on pronunciation rules. • A data-driven approach is based on a comparison between the canonical transcription and the transcription given by a phonetic recognizer.…”

Section: Introductionmentioning

confidence: 99%

Multilingual recognition of non-native speech using acoustic model transformation and pronunciation modeling

Bouselmi

Fohr

Illina

2012

Int J Speech Technol

View full text Add to dashboard Cite

This article presents an approach for the automatic recognition of non-native speech. Some non-native speakers tend to pronounce phonemes as they would in their native language. Model adaptation can improve the recognition rate for non-native speakers, but has difficulties dealing with pronunciation errors like phoneme insertions or substitutions. For these pronunciation mismatches, pronunciation modeling can make the recognition system more robust. Our approach is based on acoustic model transformation and pronunciation modeling for multiple non-native accents. For acoustic model transformation, two approaches are evaluated: MAP and model re-estimation. For pronunciation modeling, confusion rules (alternate pronunciations) are automatically extracted from a small non-native speech corpus. This paper presents a novel approach to introduce confusion rules in the recognition system which are automatically learned through pronunciation modelling. The modified HMM of a foreign spoken language phoneme includes its canonical pronunciation along with all the alternate non-native pronunciations, so that spoken language phonemes pronounced correctly by a non-native speaker could be recognized. We evaluate our approaches on the European project HIWIRE non-native corpus which contains English sentences pronounced by French, Italian, Greek and Spanish speakers. Two cases are studied: the native language of the test speaker is either known or unknown. Our approach gives better recognition results than the classical acoustic adaptation of HMM when the foreign origin of the speaker is known. We obtain 22% WER reduction compared to the reference system. Furthermore, we take into account the written form of the spoken words: non-native speakers may rely on the writing of the words in order to pronounce them. This approach does not provide any further improvements.

show abstract

On using units trained on foreign data for improved multiple accent speech recognition

Cited by 15 publications

References 24 publications

Improving proper name recognition by means of automatically learned pronunciation variants

Improving proper name recognition by means of automatically learned pronunciation variants

Lexical Modeling for Proper name Recognition in Autonomata Too

Multilingual recognition of non-native speech using acoustic model transformation and pronunciation modeling

Contact Info

Product

Resources

About