Multilingual acoustic models for speech recognition and synthesis

Kunzmann, Siegfried; Fischer, Volker; Gonzalez, Jorge; Emam, Ossama; Günther, Carsten; Janke, Eric

doi:10.1109/icassp.2004.1326652

Cited by 5 publications

(2 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In that case one can try to create acoustic models for all the sounds in the IPA (International Phonetic Alphabet) and use these models for the mapping of foreign phonemes to symbols that have an associated acoustic model (e.g. [7]). …”

Section: Introductionmentioning

confidence: 99%

Dealing with cross-lingual aspects in spoken name recognition

Stouten

Martens

2007

2007 IEEE Workshop on Automatic Speech Recognition &Amp; Understanding (ASRU)

View full text Add to dashboard Cite

The development of an automatic speech recognizer (ASR) that can accurately recognize spoken names belonging to a large lexicon, is still a big challenge. One of the bottlenecks is that many names contain elements of a foreign language origin, and native speakers can adopt very different pronunciations of these elements, ranging from completely nativized to completelyforeignized pronunciations. In this paper we further develop a recently proposed method for improving the recognition of foreign proper names spoken by native speakers. The main idea is to combine the standard acoustic model scores with scores emerging from a phonologically inspired back-off model that was trained on native speech only. This means that the proposed method does not require the development of any foreign phoneme models on foreign speech data. By applying our method on a baseline Dutch recognizer (comprising Dutch acoustic models) we could reduce the name error rate for French and English names by a considerable amount

show abstract

Section: Introductionmentioning

confidence: 99%

Dealing with cross-lingual aspects in spoken name recognition

Stouten

Martens

2007

2007 IEEE Workshop on Automatic Speech Recognition &Amp; Understanding (ASRU)

View full text Add to dashboard Cite

show abstract

“…There is also research on the use of multilingual models using statistical data-driven models known as Hidden Markov Models (HMMs). The purpose is to develop standard acoustic models of phonemes where the similarities of sounds between languages are explored [4], [5], [6]. Another possibility is to include pronunciation variants to the lexicon of the recognizer using acoustic model interpolation [7], where each model of a native-speech recognizer is interpolated with the same model from a second recognizer which depends on the speaker's accent.…”

Section: Approach and Related Workmentioning

confidence: 99%

European Portuguese Accent in Acoustic Models for Non-native English Speakers

Simões

Teixeira

Dias

et al.

Lecture Notes in Computer Science

View full text Add to dashboard Cite

The development of automatic speech recognition systems poses several known difficulties. One of them concerns the recognizer's accuracy when dealing with non-native speakers of a given language. Normally a recognizer precision is lower for non-native users, hence our goal is to improve this low accuracy rate when the speech recognition system is confronted with a foreign accent. A typical usage scenario is to apply these models in applications where European Portuguese is dominant, but where English may also frequently occur. Therefore, several experiments were performed using crossword triphone based models, which were then trained with speech corpora containing European Portuguese native speakers, English native speakers and English spoken by European Portuguese native speakers.

show abstract