Modeling letter-to-phoneme conversion as a phrase based statistical machine translation problem with minimum error rate training

Rama, Taraka; Singh, Anil Kumar; Kolachina, Sudheer

doi:10.3115/1620932.1620948

Cited by 19 publications

(8 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, this kind of writing for a word imposes challenges in the processing of the text. Letter-to-Phoneme (L2P) approach is suggested in many kinds of literature for solving lingos and slangs [29][30][31]. The CMU Pronouncing Dictionary (CMUDict) [32] is a machine-readable pronunciation dictionary contains over 125,000 words and their transcriptions.…”

Section: ) Lingoes and Slangsmentioning

confidence: 99%

Challenges with Sentiment Analysis of On-line Micro-texts

Srivastava¹,

Bhatia²

2017

IJISA

View full text Add to dashboard Cite

With the evolution of World Wide Web (WWW) 2.0 and the emergence of many micro-blogging and social networking sites like Twitter, the internet has become a massive source of short textual messages called on-line micro-texts, which are limited to a few number of characters (e.g. 140 characters on Twitter). These on-line micro-texts are considered as real-time text streams. Online micro-texts are extremely subjective; they contain opinions about various events, social issues, personalities, and products. However, despite being so voluminous in quantity, the qualitative nature of these micro-texts is very inconsistent. These qualitative inconsistencies of raw on-line micro-texts impose many challenges in sentiment analysis of on-line micro-texts by using the established methods of sentiment analysis of unstructured reviews. This paper presents many challenges and issues observed during sentiment analysis of On-line Microtexts.

show abstract

Section: ) Lingoes and Slangsmentioning

confidence: 99%

Challenges with Sentiment Analysis of On-line Micro-texts

Srivastava¹,

Bhatia²

2017

IJISA

View full text Add to dashboard Cite

show abstract

“…In this equation, an input sequence seq(g i , x) is constructed by concatenating the focal grapheme g i with its left and right context information, so the length of this sequence is equal to (2x + 1). On the other hand, considering the correspondence between graphemes and phonemes as many-to-many has also been stated as a beneficial technique in many recent studies because it can cover all possible mappings between graphemes and phonemes (e.g., one-to-one, many-toone, one-to-many, and many-to-many) [10], [11], [13], [19]. These techniques inspired us to incorporate the contextdependent phoneme model into neural network-based G2P conversion.…”

Section: Mapping Technique Between Graphemes and Phonemesmentioning

confidence: 99%

“…This corpus, which contains many acronyms and loan words from different languages such as Japanese, French, and German, has been widely used by researchers [11]- [13]. It was originally created using 34 graphemic symbols (e.g., "A".…”

Section: Auto-aligned Cmudict Corpusmentioning

confidence: 99%

“…Various many-tomany mapping techniques between letters and phonemes for taking the G2P conversion to the next level have subsequently been proposed. For example, Rama et al treat the letter-to-phoneme conversion problem as a phrase-based statistical machine translation problem [11]. They removed the one-to-one alignments from one of the most complex American English words-based dictionary (known as the auto-aligned CMUDict corpus [1]) and induced again manyto-many alignments between letters and phonemes using GIZA++ toolkit.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

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

Kheang

Katsurada

Iribe

et al. 2014

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

SUMMARYTo 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 improve the performance of G2P conversion, this paper deals with the problem of conflicting phonemes, where an input grapheme can, in the same context, produce many possible output phonemes at the same time. To this end, we propose a two-stage neural network-based approach that converts the input text to phoneme sequences in the first stage and then predicts each output phoneme in the second stage using the phonemic information obtained. The first-stage neural network is fundamentally implemented as a many-to-many mapping model for automatic conversion of word to phoneme sequences, while the second stage uses a combination of the obtained phoneme sequences to predict the output phoneme corresponding to each input grapheme in a given word. We evaluate the performance of this approach using the American English words-based pronunciation dictionary known as the auto-aligned CMUDict corpus [1]. In terms of phoneme and word accuracy of the OOV words, on comparison with several proposed baseline approaches, the evaluation results show that our proposed approach improves on the previous one-stage neural network-based approach for G2P conversion. The results of comparison with another existing approach indicate that it provides higher phoneme accuracy but lower word accuracy on a general dataset, and slightly higher phoneme and word accuracy on a selection of words consisting of more than one phoneme conflicts. key words: two-stage neural network, grapheme-to-phoneme conversion, many-to-many mapping, prediction through phonemic information, phoneme conflict

show abstract

“…As a consequence, various many-to-many mapping techniques between letters and phonemes have been proposed subsequently, in order to improve the accuracy of G2P conversion. For example, Rama et al treated the letter-tosound conversion problems as a phrase-based statistical machine translation problem [13]. The hidden Markov model (HMM)-based approach with contextsensitive observations for G2P conversion [14], proposed by Ogbureke et al, obtained a word accuracy of 79.79% on the Unilex corpora containing the UK English words, but only a maximum of 57.85% for the CMUDict corpus [15] due to the large number of loan words and a few remarkable errors.…”

Section: Related Workmentioning

confidence: 99%

New Grapheme Generation Rules for Two-Stage Modelbased Grapheme-to-Phoneme Conversion

et al. 2014

View full text Add to dashboard Cite

The precise conversion of arbitrary text into its corresponding phoneme sequence (grapheme-to-phoneme or G2P conversion) is implemented in speech synthesis and recognition, pronunciation learning software, spoken term detection and spoken document retrieval systems. Because the quality of this module plays an important role in the performance of such systems and many problems regarding G2P conversion have been reported, we propose a novel two-stage model-based approach, which is implemented using an existing weighted finite-state transducer-based G2P conversion framework, to improve the performance of the G2P conversion model. The first-stage model is built for automatic conversion of words to phonemes, while the second-stage model utilizes the input graphemes and output phonemes obtained from the first stage to determine the best final output phoneme sequence. Additionally, we designed new grapheme generation rules, which enable extra detail for the vowel and consonant graphemes appearing within a word. When compared with previous approaches, the evaluation results indicate that our approach using rules focusing on the vowel graphemes slightly improved the accuracy of the out-of-vocabulary dataset and consistently increased the accuracy of the in-vocabulary dataset.

show abstract

Modeling letter-to-phoneme conversion as a phrase based statistical machine translation problem with minimum error rate training

Cited by 19 publications

References 22 publications

Challenges with Sentiment Analysis of On-line Micro-texts

Challenges with Sentiment Analysis of On-line Micro-texts

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

New Grapheme Generation Rules for Two-Stage Modelbased Grapheme-to-Phoneme Conversion

Contact Info

Product

Resources

About