4 Hits in 6.5 sec

Multitask Learning for Grapheme-to-Phoneme Conversion of Anglicisms in German Speech Recognition [article]

Julia Pritzen, Michael Gref, Dietlind Zühlke, Christoph Schmidt
2021 arXiv   pre-print
Loanwords, such as Anglicisms, are a challenge in German speech recognition.  ...  In this work, we propose a multitask sequence-to-sequence approach for grapheme-to-phoneme conversion to improve the phonetization of Anglicisms.  ...  German G2P models show higher error rates for Anglicisms due to their irregular pronunciation compared to native German words [3] .  ... 
arXiv:2105.12708v2 fatcat:fcs3xn3v2zbzxkgwd6qjrtaetq

Introducing nativization to Spanish TTS systems

Tatyana Polyákova, Antonio Bonafonte
2011 Speech Communication  
Mass media globalization introduces multilingualism as a challenge for the most popular speech applications such as text-to-speech synthesis and automatic speech recognition.  ...  In Spain and other Spanish-speaking countries, the use of Anglicisms and other words of foreign origin is constantly growing.  ...  of letters, being that G2P conversion is already a difficult task for English.  ... 
doi:10.1016/j.specom.2011.05.009 fatcat:3uj5kdx4i5bqdp7nfqkur7woua

Rapid Generation of Pronunciation Dictionaries for new Domains and Languages

Tim Schlippe
Starting from the straightforward scenario in which the target language is present in written form on the Internet and the mapping between speech and written language is close up to the difficult scenario  ...  This dissertation presents innovative strategies and methods for the rapid generation of pronunciation dictionaries for new domains and languages.  ...  She believed in my research and supported me with many useful discussions. Her great personality and excellent research skills had a very strong effect on my scientific career.  ... 
doi:10.5445/ir/1000044928 fatcat:26gf3xyoz5cbvduozxce4gprze

Improving spoken document retrieval by unsupervised language model adaptation using utterance-based web search

Robert Herms, Marc Ritter, Thomas Wilhelm-Stein, Maximilian Eibl
2014 Interspeech 2014   unpublished
These data are used to perform a block-specific adaptation of a general pronunciation dictionary and a general LM.  ...  Our experimental results show improvements of up to 11.7% for MAP of 18 different topics and 7.5% of WER in comparison to the base LM.  ...  Acknowledgements This work was realized as part of the project Chrooma+ supported by the Sächsische Aufbaubank within the European Social Fund in the Free State of Saxony, Germany and the project ValidAX  ... 
doi:10.21437/interspeech.2014-350 fatcat:rxnqxgtqpreqjlsj5pto6lhwee