Filters








3,761 Hits in 4.2 sec

Memory-Based Phoneme-to-Grapheme Conversion: A Method for Dealing with Out-of-Vocabulary Items in Speech Recognition [chapter]

Bart Decadt, Jacques Duchateau, Walter Daelemans, Patrick Wambacq
Computational Linguistics in the Netherlands 2001  
The basic idea is to indicate uncertain words in the transcriptions and replace them with phoneme recognition results that are post-processed using a phoneme-to-grapheme (P2G) converter.  ...  We concentrate on the final step, P2G conversion: we show that the phoneme recognition results can be reasonably reliably transcribed orthographically using machine learning techniques.  ...  Acknowledgments This research is funded by IWT in the STWW programme, project ATraNoS. 5 We would like to thank Erik Tjong Kim Sang and Véronique Hoste from the CNTS research group for their support  ... 
doi:10.1163/9789004334038_006 fatcat:nvebm6dzuzfejlpn5clphgyctm

Transcription of out-of-vocabulary words in large vocabulary speech recognition based on phoneme-to-grapheme conversion

Decadt, Duchateau, Daelemans, Wambacq
2002 IEEE International Conference on Acoustics Speech and Signal Processing  
The basic idea is to replace uncertain words in the transcriptions with a phoneme recognition result that is postprocessed using a phoneme-to-grapheme converter.  ...  This converter turns phoneme strings into grapheme strings and is trained using machine learning techniques. Recently, the system was enhanced by adding a spelling checker to it.  ...  The use of a phoneme graph, possibly including probabilities for the phonemes, could be a solution to this problem.  ... 
doi:10.1109/icassp.2002.1005876 fatcat:cagcqc5mf5dtrbpbtt2vjzzacu

Transcription of out-of-vocabulary words in large vocabulary speech recognition based on phoneme-to-grapheme conversion

Bart Decadt, Jacques Duchateau, Walter Daelemans, Patrick Wambacq
2002 IEEE International Conference on Acoustics Speech and Signal Processing  
The basic idea is to replace uncertain words in the transcriptions with a phoneme recognition result that is postprocessed using a phoneme-to-grapheme converter.  ...  This converter turns phoneme strings into grapheme strings and is trained using machine learning techniques. Recently, the system was enhanced by adding a spelling checker to it.  ...  The use of a phoneme graph, possibly including probabilities for the phonemes, could be a solution to this problem.  ... 
doi:10.1109/icassp.2002.5743875 dblp:conf/icassp/DecadtDDW02 fatcat:xzjsbkcgrjcxng43vwe2lipkje

Persian sentences to phoneme sequences conversion based on recurrent neural networks

Yasser Mohseni Behbahani, Bagher Babaali, Mussa Turdalyuly
2016 Open Computer Science  
AbstractGrapheme to phoneme conversion is one of the main subsystems of Text-to-Speech (TTS) systems.  ...  In this paper we define the grapheme-to-phoneme conversion as a sequential labeling problem; and use the modified Recurrent Neural Networks (RNN) to create a smart and integrated model for this purpose  ...  Acknowledgement: The authors would like to thank the NVIDIA Company for the donated GPU model: Tesla K40. All experiments in this paper are conducted using this GPU.  ... 
doi:10.1515/comp-2016-0019 fatcat:begsp4btnjhnvbk36iy25gplwi

Grapheme-to-phoneme conversion using Long Short-Term Memory recurrent neural networks

Kanishka Rao, Fuchun Peng, Hasim Sak, Francoise Beaufays
2015 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)  
conversions to a word-to-pronunciation conversion.  ...  Training joint-sequence based G2P require explicit graphemeto-phoneme alignments which are not straightforward since graphemes and phonemes don't correspond one-to-one.  ...  CONCLUSION We suggested LSTM-based architectures to perform G2P conversions.  ... 
doi:10.1109/icassp.2015.7178767 dblp:conf/icassp/RaoPSB15 fatcat:rzn5y3jd3jbghknshsktk24wee

Deep Bidirectional Long Short-Term Memory Recurrent Neural Networks for Grapheme-to-Phoneme Conversion Utilizing Complex Many-to-Many Alignments

Amr El-Desoky Mousa, Björn Schuller
2016 Interspeech 2016  
Efficient grapheme-to-phoneme (G2P) conversion models are considered indispensable components to achieve the stateof-the-art performance in modern automatic speech recognition (ASR) and text-to-speech  ...  Recent work in this domain is based on recurrent neural networks (RNN) that are capable of translating grapheme sequences into phoneme sequences taking into account the full context of graphemes.  ...  Successful approaches to G2P conversion use joint sequence models [1, 2] . Therein, an initial grapheme-phoneme sequence alignment is created.  ... 
doi:10.21437/interspeech.2016-1229 dblp:conf/interspeech/MousaS16 fatcat:adglauvevfbdbdbmncvohvokp4

Treetalk: Memory-Based Word Phonemisation [chapter]

Walter Daelemans, Antal van den Bosch
2001 Data-Driven Techniques in Speech Synthesis  
We propose a memory-based (similarity-based) approach to learning the mapping of words into phonetic representations for use in speech synthesis systems.  ...  TRIBL was used in TREETALK, a methodology for fast engineering of word-to-phonetics conversion systems.  ...  M-G-S The subtasks of graphemic parsing (A) and grapheme-phoneme conversion (G) are clearly related. While A attempts to parse a letter string into graphemes, G converts graphemes to phonemes.  ... 
doi:10.1007/978-1-4757-3413-3_7 fatcat:omp5zcjb7rbnjghlmhxrzh4h6q

An analysis of writing in a case of deep dyslexia

Karen A. Nolan, Alfonso Caramazza
1983 Brain and Language  
A model of writing is proposed which explains these errors in terms of a disruption of a phoneme-grapheme conversion process which normally functions to prevent decay of information from a Graphemic Buffer  ...  and abstractness effects in oral reading, oral and written naming, and writing to dictation, but not in repetition of single words and copying from memory.  ...  The research reported in this paper was supported by National Institute of Health (NINCDS) Grant NS-16155 to the Johns Hopkins University. Send requests for reprints to Dr. Karen A.  ... 
doi:10.1016/0093-934x(83)90047-0 pmid:6640282 fatcat:rijyokq2ffb7nhzkd475qk4mqi

An analysis of writing in a case of deep dyslexia

K. A. Nolan
2000 Neurocase  
A model of writing is proposed which explains these errors in terms of a disruption of a phoneme-grapheme conversion process which normally functions to prevent decay of information from a Graphemic Buffer  ...  and abstractness effects in oral reading, oral and written naming, and writing to dictation, but not in repetition of single words and copying from memory.  ...  The research reported in this paper was supported by National Institute of Health (NINCDS) Grant NS-16155 to the Johns Hopkins University. Send requests for reprints to Dr. Karen A.  ... 
doi:10.1093/neucas/6.2.164 fatcat:zcpcrxpqufeatgfs6nkpdaqdcq

Solving the Phoneme Conflict in Grapheme-to-Phoneme Conversion Using a Two-Stage Neural Network-Based Approach

Seng KHEANG, Kouichi KATSURADA, Yurie IRIBE, Tsuneo NITTA
2014 IEICE transactions on information and systems  
neural network-based approach for G2P conversion.  ...  To achieve high quality output speech synthesis systems, data-driven grapheme-to-phoneme (G2P) conversion is usually used to generate the phonetic transcription of out-of-vocabulary (OOV) words.  ...  To output the phonemes of the input text, prediction must be based on phonemic rather than graphemic information.  ... 
doi:10.1587/transinf.e97.d.901 fatcat:6eddqluzczf73jotfmplonvbau

A Finite State and Data-Oriented Method for Grapheme to Phoneme Conversion [article]

Gosse Bouma
2000 arXiv   pre-print
A finite-state method, based on leftmost longest-match replacement, is presented for segmenting words into graphemes, and for converting graphemes into phonemes.  ...  A small set of hand-crafted conversion rules for Dutch achieves a phoneme accuracy of over 93%. The accuracy of the system is further improved by using transformation-based learning.  ...  Figure 2: Conversion Rules between graphemes and phonemes is usually one to one, but it is no problem to align a grapheme with two or more phonemes.  ... 
arXiv:cs/0003074v1 fatcat:heyzhdvo3fcdfklm5xhgybzjp4

A machine transliteration model based on correspondence between graphemes and phonemes

Jong-Hoon Oh, Key-Sun Choi, Hitoshi Isahara
2006 ACM Transactions on Asian Language Information Processing  
Three types of machine transliteration models-grapheme-based, phoneme-based, and hybrid-have been proposed.  ...  Furthermore, little work has been reported on ways to dynamically handle source graphemes and phonemes.  ...  The phonemes are then transformed into Korean PUs using the "English-to-Korean Standard Conversion Rules" (EKSCRS), which describe the conversion of English phonemes into Korean graphemes using the phonemes  ... 
doi:10.1145/1194936.1194938 fatcat:vnaz3ca2wbbtzokxzelmj32bme

Letter-to-Sound Pronunciation Prediction Using Conditional Random Fields

Dong Wang, Simon King
2011 IEEE Signal Processing Letters  
One challenge in applying CRFs to LTS is that the phoneme and grapheme sequences of a word are generally of different lengths, which makes CRF training difficult.  ...  Pronunciation prediction, or letter-to-sound (LTS) conversion, is an essential task for speech synthesis, open vocabulary spoken term detection and other applications dealing with novel words.  ...  First, features can be defined easily by specifying the concerned graphemes and phonemes; second, the limited memory BFGS (LBFGS) algorithm makes the training fast with moderate memory usage, allowing  ... 
doi:10.1109/lsp.2010.2098440 fatcat:m3qzrrcgdjgbdmiqr4rcmhg54i

Successful treatment of sublexical reading deficits in a child with dyslexia of the mixed type

Ruth K. Brunsdon, Timothy J. Hannan, Lyndsey Nickels, Max Coltheart
2002 Neuropsychological Rehabilitation  
The sublexical reading procedure involves rule-based grapheme-to-phoneme conversion and allows the skilled reader to "sound-out" unfamiliar words and nonwords.  ...  Second, each identified grapheme must be assigned the correct sound or phoneme using learned and stored grapheme-to-phoneme conversion rules.  ... 
doi:10.1080/09602010244000048 fatcat:ndgzhlxyyrgezb7kw7ynca34ki

Letter-to-sound conversion using coupled Hidden Markov Models for lexicon compression

Hao Che, Jianhua Tao, Shifeng Pan
2012 2012 International Conference on Speech Database and Assessments  
Two Hidden Markov Models (HMM) which are respectively designed to predict the best phonemic string and corresponding graphemic substring segmentation are coupled in the phase of phonemes generation.  ...  Letter-to-Sound(LTS) conversion, which is used to compress the lexicon for embedded application purpose, has become an important part in Text-to-Speech (TTS) system.  ...  based on one-to-one alignment.  ... 
doi:10.1109/icsda.2012.6422464 fatcat:ymoxstz3sbbepilbl55kxtwx2a
« Previous Showing results 1 — 15 out of 3,761 results