An Attentional Model for Speech Translation Without Transcription
Published 1 January 2016Open access
Long Duong, Antonios Anastasopoulos, David Chiang, Steven Bird, Trevor Cohn
Citations160
Generate an AI Snapshot to get a quick, structured summary of this paper.
Study Snapshot
ObjectiveStudy objective
MethodsResearch methodology
PopulationPopulation studied
Sample sizeSample sizes
OutcomesStudy outcomes here
ResultsStudy results comes here
LimitationsResearch study limitations comes here
A concise AI-generated summary of the paper will appear here once you click Generate AI Snapshot.
TL;DR
On the more challenging speech-to-word alignment task, the model nearly matches GIZA++’s performance on gold transcriptions, but without recourse to transcriptions or to a lexicon.
Abstract
Long Duong, Antonios Anastasopoulos, David Chiang, Steven Bird, Trevor Cohn. Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2016.
Keywords
Computer Science
Neural ComputationLong Short-Term Memory
98,079 Citations1997Sepp Hochreiter, Jürgen Schmidhuber
A novel, efficient, gradient based method called long short-term memory (LSTM) is introduced, which can learn to bridge minimal time lags in excess of 1000 discrete-time steps by enforcing constant error flow through constant error carousels within special units.
Dropout: a simple way to prevent neural networks from overfitting
34,279 Citations2014Nitish Srivastava, Geoffrey E. Hinton +3 more
It is shown that dropout improves the performance of neural networks on supervised learning tasks in vision, speech recognition, document classification and computational biology, obtaining state-of-the-art results on many benchmark data sets.
arXiv (Cornell University)Neural Machine Translation by Jointly Learning to Align and Translate
14,565 Citations2014Dzmitry Bahdanau
arXiv (Cornell University)Sequence to Sequence Learning with Neural Networks
13,362 Citations2014Ilya Sutskever, Oriol Vinyals +1 more
Speech recognition with deep recurrent neural networks
8,837 Citations2013Alex Graves, Abdelrahman Mohamed +1 more
This paper investigates deep recurrent neural networks, which combine the multiple levels of representation that have proved so effective in deep networks with the flexible use of long range context that empowers RNNs.
Adaptive Subgradient Methods for Online Learning and Stochastic Optimization
8,612 Citations2011John C. Duchi, Elad Hazan +1 more
Effective Approaches to Attention-based Neural Machine Translation
8,582 Citations2015Thang Luong, Hieu Pham +1 more
A global approach which always attends to all source words and a local one that only looks at a subset of source words at a time are examined, demonstrating the effectiveness of both approaches on the WMT translation tasks between English and German in both directions.
arXiv (Cornell University)Show, Attend and Tell: Neural Image Caption Generation with Visual Attention
7,525 Citations2015Kelvin Xu, Jimmy Ba +6 more
An attention based model that automatically learns to describe the content of images is introduced that can be trained in a deterministic manner using standard backpropagation techniques and stochastically by maximizing a variational lower bound.
arXiv (Cornell University)ADADELTA: An Adaptive Learning Rate Method
5,531 Citations2012Matthew D. Zeiler
A novel per-dimension learning rate method for gradient descent called ADADELTA that dynamically adapts over time using only first order information and has minimal computational overhead beyond vanilla stochastic gradient descent is presented.
Connectionist temporal classification
5,299 Citations2006Alex Graves, Santiago Fernández +2 more
Moses
4,868 Citations2007Philipp Koehn, Richard Zens +12 more
An open-source toolkit for statistical machine translation whose novel contributions are support for linguistically motivated factors, confusion network decoding, and efficient data formats for translation models and language models.
The mathematics of statistical machine translation: parameter estimation
4,125 Citations1993Peter F. Brown, Vincent J. Della Pietra +2 more
It is reasonable to argue that word-by-word alignments are inherent in any sufficiently large bilingual corpus, given a set of pairs of sentences that are translations of one another.
The Journal of the Acoustical Society of AmericaPerceptual linear predictive (PLP) analysis of speech
2,539 Citations1990Hynek Heřmanský
A new technique for the analysis of speech, the perceptual linear predictive (PLP) technique, which uses three concepts from the psychophysics of hearing to derive an estimate of the auditory spectrum, and yields a low-dimensional representation of speech.
Listen, attend and spell: A neural network for large vocabulary conversational speech recognition
2,308 Citations2016William Chan, Navdeep Jaitly +2 more
Listen, Attend and Spell (LAS), a neural speech recognizer that transcribes speech utterances directly to characters without pronunciation models, HMMs or other components of traditional speech recognizers is presented.
arXiv (Cornell University)Attention-Based Models for Speech Recognition
1,780 Citations2015Jan Chorowski, Dzmitry Bahdanau +3 more
The attention-mechanism is extended with features needed for speech recognition and a novel and generic method of adding location-awareness to the attention mechanism is proposed to alleviate the issue of high phoneme error rate.
arXiv (Cornell University)Show, Attend and Tell: Neural Image Caption Generation with Visual\n Attention
1,754 Citations2015Kelvin Xu, Jimmy Ba +6 more
RWTH Publications (RWTH Aachen)Moses: Open Source Toolkit for Statistical Machine Translation
1,451 Citations2007Philipp Koehn, Hieu Hoang +12 more
Improved statistical alignment models
1,015 Citations2000Franz Josef Och, Hermann Ney
It is shown that models with a first-order dependence and a fertility model lead to significantly better results than the simple models IBM-1 or IBM-2, which are not able to go beyond zero-order dependencies.
Research Showcase @ Carnegie Mellon University (Carnegie Mellon University)A Simple, Fast, and Effective Reparameterization of IBM Model 2
839 Citations2018Chris Dyer, Victor Chahuneau +1 more
A simple log-linear reparameterization of IBM Model 2 that overcomes problems arising from Model 1’'s strong assumptions and Model 2’s overparameterization is presented.
HMM-based word alignment in statistical translation
829 Citations1996Stephan Vogel, Hermann Ney +1 more
A new model for word alignment in statistical translation using a first-order Hidden Markov model for the word alignment problem as they are used successfully in speech recognition for the time alignment problem.
WIT3: Web Inventory of Transcribed and Translated Talks
579 Citations2012Mauro Cettolo, Christian Girardi +1 more
A Web inventory named WIT 3 that offers access to a collection of transcribed and translated talks from the TED Talks corpus, that basically redistributes the original content published by the TED Conference website.
arXiv (Cornell University)End-to-end Continuous Speech Recognition using Attention-based Recurrent NN: First Results
415 Citations2014Jan Chorowski, Dzmitry Bahdanau +2 more
Initial results demonstrate that this new approach achieves phoneme error rates that are comparable to the state-of-the-art HMM-based decoders, on the TIMIT dataset.
Canadian Acoustical Association - Canadian Acoustics: Journal of the Canadian Acoustical AssociationProsodylab-aligner: A tool for forced alignment of laboratory speech
228 Citations2011Kyle Gorman, Jonathan Howell +1 more
The Penn Forced Aligner automates the alignment process using the Hidden Markov Model Toolkit (HTK), and the core of Prosodylab-Aligner is align, a script which performs acoustic model training and alignment.
Advances in Engineering SoftwareAnalysis of learning rate and momentum term in backpropagation neural network algorithm trained to predict pavement performance
156 Citations1999Nii Attoh-Okine
Real pavement condition and traffic data and specific architecture are used to investigate the effect of learning rate and momentum term on back-propagation algorithm neural network trained to predict flexible pavement performance and it is concluded that an extremely low learning rate does not give satisfactory results for the specific data set and the architecture used.
Lexicon-Free Conversational Speech Recognition with Neural Networks
129 Citations2015Andrew L. Maas, Ziang Xie +2 more
An approach to speech recognition that uses only a neural network to map acoustic input to characters, a character-level language model, and a beam search decoding procedure, making it possible to directly train a speech recognizer using errors generated by spoken language understanding tasks.
Edinburgh Research Explorer (University of Edinburgh)Improved Speech-to-Text Translation with the Fisher and Callhome Spanish–English Speech Translation Corpus
110 Citations2013Matt Post, Gaurav Kumar +4 more
The Fisher and Callhome Spanish-English Speech Translation Corpus is introduced, supplementing existing LDC audio and transcripts with ASR 1-best, lattice, and oracle output produced by the Kaldi recognition system and English translations obtained on Amazon’s Mechanical Turk.
An Unsupervised Model for Joint Phrase Alignment and Extraction
71 Citations2011Graham Neubig, Taro Watanabe +3 more
An unsupervised model for joint phrase alignment and extraction using non-parametric Bayesian methods and inversion transduction grammars (ITGs) is presented, which matches the accuracy of traditional two-step word alignment/phrase extraction approach while reducing the phrase table to a fraction of the original size.
Minerva Access (University of Melbourne)The Human Language Project: Building a Universal Corpus of the World's Languages
60 Citations2010Steven Abney, Steven Bird
It is proposed that the ability to train systems to translate into and out of a given language be the yardstick for determining when the authors have successfully captured a language.
Joint Learning of Phonetic Units and Word Pronunciations for ASR
39 Citations2013Chia-ying Lee, Yu Zhang +1 more
An unsupervised alternative ‐ requiring no language-specific knowledge ‐ to the conventional manual approach for creating pronunciation dictionaries is proposed, which jointly discovers the phonetic inventory and the Letter-to-Sound mapping rules in a language using only transcribed data.
Machine Translation without Words through Substring Alignment
37 Citations2012Graham Neubig, Taro Watanabe +2 more
This paper demonstrates that accurate machine translation is possible without the concept of "words," treating MT as a problem of transformation between character strings, and proposes a look-ahead parsing algorithm and substring-informed prior probabilities to achieve more effective and efficient alignment.
Collecting Bilingual Audio in Remote Indigenous Communities
28 Citations2014Steven Bird, Lauren Gawne +2 more
It is argued that the use of special-purpose software on smartphones is an effective and scalable method for large-scale collection of bilingual audio, and ultimately bilingual text, for languages spoken in remote indigenous communities.
Lecture notes in computer scienceA Scalable Method for Preserving Oral Literature from Small Languages
19 Citations2010Steven Bird
It is concluded that Basic Oral Language Documentation is a promising technique for expediting the task of preserving endangered linguistic heritage.
Word segmentation through cross-lingual word-to-phoneme alignment
17 Citations2012Felix Stahlberg, Tim Schlippe +2 more
It is shown that unsupervised learning of word segmentation is more accurate when information of another language is used, and that the new alignment model Model 3P for cross-lingual word-to-phoneme alignment outperforms a state-of-the-art monolingualword segmentation approach.
Machine Translation for Language Preservation
17 Citations2012Steven Bird, David Chiang
This position paper examines the challenges of solving the problem that for most of the languages in question, no texts or lexicons exist, and reports on a data collection effort involving 15 endangered languages spoken in the highlands of Papua New Guinea.
arXiv (Cornell University)Sequence Modeling using Gated Recurrent Neural Networks
9 Citations2015Mohammad Pezeshki
This paper uses Recurrent Neural Networks to capture and model human motion data and generate motions by prediction of the next immediate data point at each time-step and demonstrates that this model is able to capture long-term dependencies in data and generated realistic motions.
Signals and communication technologyHierarchical Neural Network Structures for Phoneme Recognition
8 Citations2012Daniel Vásquez, Rainer Gruhn +1 more
In this book, hierarchical structures based on neural networks are investigated for automatic speech recognition under the Hybrid Hidden Markov Model/Artificial Neural Network (HMM/ANN) paradigm.