Joint-sequence models for grapheme-to-phoneme conversion

By M. Bisani, Hermann Ney. 2008

801 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Natural Language Processing, Speech & Audio

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Natural Language Processing

In Speech & Audio

Citations in the corpus

Librispeech: An ASR corpus based on public domain audio books (2015, methodology): “5% of all tokens in the eva luation sets; we generated pronunciations for the remaining words u sing the Sequitur G2P toolkit [11].”
A Survey on Neural Speech Synthesis (2021, methodology): “The text analysis module first processes the text, including text normalization [311], grapheme-to-phoneme conversion [24], word segmentation, etc, and then extracts the linguistic features, such as phonemes, duration and POS tags from different granularities.”
Sequence-to-sequence neural net models for grapheme-to-phoneme conversion (2015, methodology): “Joint modeling has been proposed for grapheme-tophoneme conversion [18, 19, 21].”
Attention with Intention for a Neural Network Conversation Model (2015, background): “For example, the alignment information between the source and target side is critical in grapheme-to-phoneme conversation [17] to outperform a strong baseline using n-gram models [2].”
Token-Level Ensemble Distillation for Grapheme-to-Phoneme Conversion (2019, methodology): “To be consistent with the previous works [4, 5, 21], stress markings are removed and the multiple pronunciations are retained.”
Knowledge Distillation from Bert in Pre-Training and Fine-Tuning for Polyphone Disambiguation (2019, background): “For alphabetic languages such as English and French, the main challenge of G2P conversion is how to predict the pronunciations of out-of-vocabulary (OOV) words [3, 4, 5].”
Sentence-level control vectors for deep neural network speech synthesis (2015, methodology): “A letter-to-sound predictor based on joint multigrams [23] was trained on this lexicon to handle out-of-vocabulary words.”
State-of-the-Art Speech Recognition Using Multi-Stream Self-Attention with Dilated 1D Convolutions (2019, methodology): “Using the base lexicon of the CMUdict 4 that covered 81K among the selected words of 522K, we trained a G2P model using the Sequitur tool [26] to cover the out-of-vocabulary words.”
Text-to-Speech Synthesis Using Found Data for Low-Resource Languages (2019, methodology): “Pronunciations for out-of-vocabulary words (words contained in the broadcast news transcripts but not in the BABEL lexicon) were obtained by training a pronunciation model on the original lexicon using Sequitur G2P (Bisani and Ney, 2008) and then using the trained model to generate phonetic sequence”
From Senones to Chenones: Tied Context-Dependent Graphemes for Hybrid Speech Recognition (2019, background): “We can further train a g2p model on these lexicons to predict pronunciations for previously unseen words [31].”
A bi-directional LSTM approach for polyphone disambiguation in Mandarin Chinese (2016, background): “In the G2P conversion task [1, 2, 3, 4, 5, 6, 7], a neural network accepts a sequence of characters and outputs a sequence of pronunciations.”
Grapheme-to-phoneme conversion using Long Short-Term Memory recurrent neural networks (2015, methodology): “In such a formulation, an alignment model needs to be first constructed and then a translation model – such as a joint ngram model – is built from the alignments [2, 4].”
Polyphone Disambiguation for Mandarin Chinese Using Conditional Neural Network with Multi-level Embedding Features (2019, methodology): “While the G2P system in English TTS synthesis system aims to produce the phoneme sequences for the out-of-lexicon words, The target of a G2P system in Chinese TTS synthesis system is to convert Chinese characters to pinyins (phoneme representations with Latin alphabet in Mandarin Chinese) [10].”
Word-Level Speech Recognition With a Letter to Word Encoder (2019, background): “Traditional speech recognizers also have adaptable lexicons, which they achieve by predicting phonemes and requiring a model for grapheme-to-phoneme conversion (Bisani & Ney, 2008; Rao et al., 2015).”
Confidence Estimation and Deletion Prediction Using Bidirectional Recurrent Neural Networks (2018, methodology): “Pronunciations for these words were obtained by a grapheme-to-phoneme model [35].”