Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox
Picture for Boris Ginsburg

Boris Ginsburg

Adapting TTS models For New Speakers using Transfer Learning


Oct 12, 2021
Paarth Neekhara, Jason Li, Boris Ginsburg

Figure 1 for Adapting TTS models For New Speakers using Transfer Learning
Figure 2 for Adapting TTS models For New Speakers using Transfer Learning
* Submitted to ICASSP 2022 
  Access Paper or Ask Questions

TitaNet: Neural Model for speaker representation with 1D Depth-wise separable convolutions and global context


Oct 08, 2021
Nithin Rao Koluguri, Taejin Park, Boris Ginsburg

Figure 1 for TitaNet: Neural Model for speaker representation with 1D Depth-wise separable convolutions and global context
Figure 2 for TitaNet: Neural Model for speaker representation with 1D Depth-wise separable convolutions and global context
Figure 3 for TitaNet: Neural Model for speaker representation with 1D Depth-wise separable convolutions and global context
Figure 4 for TitaNet: Neural Model for speaker representation with 1D Depth-wise separable convolutions and global context
* preprint. Submitted to ICASSP 2022 
  Access Paper or Ask Questions

Mixer-TTS: non-autoregressive, fast and compact text-to-speech model conditioned on language model embeddings


Oct 07, 2021
Oktai Tatanov, Stanislav Beliaev, Boris Ginsburg

Figure 1 for Mixer-TTS: non-autoregressive, fast and compact text-to-speech model conditioned on language model embeddings
Figure 2 for Mixer-TTS: non-autoregressive, fast and compact text-to-speech model conditioned on language model embeddings
Figure 3 for Mixer-TTS: non-autoregressive, fast and compact text-to-speech model conditioned on language model embeddings
Figure 4 for Mixer-TTS: non-autoregressive, fast and compact text-to-speech model conditioned on language model embeddings
* Preprint. Submitted to ICASSP-22 
  Access Paper or Ask Questions

CTC Variations Through New WFST Topologies


Oct 06, 2021
Aleksandr Laptev, Somshubra Majumdar, Boris Ginsburg

Figure 1 for CTC Variations Through New WFST Topologies
Figure 2 for CTC Variations Through New WFST Topologies
Figure 3 for CTC Variations Through New WFST Topologies
Figure 4 for CTC Variations Through New WFST Topologies
* Submitted to ICASSP 2022, 5 pages, 2 figures, 7 tables 
  Access Paper or Ask Questions

A Unified Transformer-based Framework for Duplex Text Normalization


Aug 23, 2021
Tuan Manh Lai, Yang Zhang, Evelina Bakhturina, Boris Ginsburg, Heng Ji

Figure 1 for A Unified Transformer-based Framework for Duplex Text Normalization
Figure 2 for A Unified Transformer-based Framework for Duplex Text Normalization
Figure 3 for A Unified Transformer-based Framework for Duplex Text Normalization
Figure 4 for A Unified Transformer-based Framework for Duplex Text Normalization
* Under Review 
  Access Paper or Ask Questions

CarneliNet: Neural Mixture Model for Automatic Speech Recognition


Jul 22, 2021
Aleksei Kalinov, Somshubra Majumdar, Jagadeesh Balam, Boris Ginsburg

Figure 1 for CarneliNet: Neural Mixture Model for Automatic Speech Recognition
Figure 2 for CarneliNet: Neural Mixture Model for Automatic Speech Recognition
Figure 3 for CarneliNet: Neural Mixture Model for Automatic Speech Recognition
Figure 4 for CarneliNet: Neural Mixture Model for Automatic Speech Recognition
* Submitted to ASRU 2021 
  Access Paper or Ask Questions

SGD-QA: Fast Schema-Guided Dialogue State Tracking for Unseen Services


May 17, 2021
Yang Zhang, Vahid Noroozi, Evelina Bakhturina, Boris Ginsburg

Figure 1 for SGD-QA: Fast Schema-Guided Dialogue State Tracking for Unseen Services
Figure 2 for SGD-QA: Fast Schema-Guided Dialogue State Tracking for Unseen Services
Figure 3 for SGD-QA: Fast Schema-Guided Dialogue State Tracking for Unseen Services
Figure 4 for SGD-QA: Fast Schema-Guided Dialogue State Tracking for Unseen Services

  Access Paper or Ask Questions

TalkNet 2: Non-Autoregressive Depth-Wise Separable Convolutional Model for Speech Synthesis with Explicit Pitch and Duration Prediction


Apr 19, 2021
Stanislav Beliaev, Boris Ginsburg

Figure 1 for TalkNet 2: Non-Autoregressive Depth-Wise Separable Convolutional Model for Speech Synthesis with Explicit Pitch and Duration Prediction
Figure 2 for TalkNet 2: Non-Autoregressive Depth-Wise Separable Convolutional Model for Speech Synthesis with Explicit Pitch and Duration Prediction
Figure 3 for TalkNet 2: Non-Autoregressive Depth-Wise Separable Convolutional Model for Speech Synthesis with Explicit Pitch and Duration Prediction
Figure 4 for TalkNet 2: Non-Autoregressive Depth-Wise Separable Convolutional Model for Speech Synthesis with Explicit Pitch and Duration Prediction
* arXiv admin note: substantial text overlap with arXiv:2005.05514 
  Access Paper or Ask Questions

NeMo Inverse Text Normalization: From Development To Production


Apr 11, 2021
Yang Zhang, Evelina Bakhturina, Kyle Gorman, Boris Ginsburg

Figure 1 for NeMo Inverse Text Normalization: From Development To Production
Figure 2 for NeMo Inverse Text Normalization: From Development To Production
Figure 3 for NeMo Inverse Text Normalization: From Development To Production
Figure 4 for NeMo Inverse Text Normalization: From Development To Production

  Access Paper or Ask Questions

NeMo Toolbox for Speech Dataset Construction


Apr 11, 2021
Evelina Bakhturina, Vitaly Lavrukhin, Boris Ginsburg

Figure 1 for NeMo Toolbox for Speech Dataset Construction
Figure 2 for NeMo Toolbox for Speech Dataset Construction
Figure 3 for NeMo Toolbox for Speech Dataset Construction
Figure 4 for NeMo Toolbox for Speech Dataset Construction

  Access Paper or Ask Questions

SPGISpeech: 5,000 hours of transcribed financial audio for fully formatted end-to-end speech recognition


Apr 06, 2021
Patrick K. O'Neill, Vitaly Lavrukhin, Somshubra Majumdar, Vahid Noroozi, Yuekai Zhang, Oleksii Kuchaiev, Jagadeesh Balam, Yuliya Dovzhenko, Keenan Freyberg, Michael D. Shulman, Boris Ginsburg, Shinji Watanabe, Georg Kucsko

Figure 1 for SPGISpeech: 5,000 hours of transcribed financial audio for fully formatted end-to-end speech recognition
Figure 2 for SPGISpeech: 5,000 hours of transcribed financial audio for fully formatted end-to-end speech recognition
Figure 3 for SPGISpeech: 5,000 hours of transcribed financial audio for fully formatted end-to-end speech recognition
Figure 4 for SPGISpeech: 5,000 hours of transcribed financial audio for fully formatted end-to-end speech recognition
* 5 pages, 1 figure. Submitted to INTERSPEECH 2021 
  Access Paper or Ask Questions

Citrinet: Closing the Gap between Non-Autoregressive and Autoregressive End-to-End Models for Automatic Speech Recognition


Apr 05, 2021
Somshubra Majumdar, Jagadeesh Balam, Oleksii Hrinchuk, Vitaly Lavrukhin, Vahid Noroozi, Boris Ginsburg

Figure 1 for Citrinet: Closing the Gap between Non-Autoregressive and Autoregressive End-to-End Models for Automatic Speech Recognition
Figure 2 for Citrinet: Closing the Gap between Non-Autoregressive and Autoregressive End-to-End Models for Automatic Speech Recognition
Figure 3 for Citrinet: Closing the Gap between Non-Autoregressive and Autoregressive End-to-End Models for Automatic Speech Recognition
Figure 4 for Citrinet: Closing the Gap between Non-Autoregressive and Autoregressive End-to-End Models for Automatic Speech Recognition

  Access Paper or Ask Questions

Hi-Fi Multi-Speaker English TTS Dataset


Apr 03, 2021
Evelina Bakhturina, Vitaly Lavrukhin, Boris Ginsburg, Yang Zhang

Figure 1 for Hi-Fi Multi-Speaker English TTS Dataset
Figure 2 for Hi-Fi Multi-Speaker English TTS Dataset
Figure 3 for Hi-Fi Multi-Speaker English TTS Dataset
Figure 4 for Hi-Fi Multi-Speaker English TTS Dataset

  Access Paper or Ask Questions

On regularization of gradient descent, layer imbalance and flat minima


Jul 18, 2020
Boris Ginsburg

Figure 1 for On regularization of gradient descent, layer imbalance and flat minima
Figure 2 for On regularization of gradient descent, layer imbalance and flat minima

  Access Paper or Ask Questions

Correction of Automatic Speech Recognition with Transformer Sequence-to-sequence Model


Oct 23, 2019
Oleksii Hrinchuk, Mariya Popova, Boris Ginsburg

Figure 1 for Correction of Automatic Speech Recognition with Transformer Sequence-to-sequence Model
Figure 2 for Correction of Automatic Speech Recognition with Transformer Sequence-to-sequence Model
Figure 3 for Correction of Automatic Speech Recognition with Transformer Sequence-to-sequence Model
Figure 4 for Correction of Automatic Speech Recognition with Transformer Sequence-to-sequence Model

  Access Paper or Ask Questions

NeMo: a toolkit for building AI applications using Neural Modules


Sep 14, 2019
Oleksii Kuchaiev, Jason Li, Huyen Nguyen, Oleksii Hrinchuk, Ryan Leary, Boris Ginsburg, Samuel Kriman, Stanislav Beliaev, Vitaly Lavrukhin, Jack Cook, Patrice Castonguay, Mariya Popova, Jocelyn Huang, Jonathan M. Cohen

Figure 1 for NeMo: a toolkit for building AI applications using Neural Modules
Figure 2 for NeMo: a toolkit for building AI applications using Neural Modules
* 6 pages plus references 
  Access Paper or Ask Questions

Stochastic Gradient Methods with Layer-wise Adaptive Moments for Training of Deep Networks


May 27, 2019
Boris Ginsburg, Patrice Castonguay, Oleksii Hrinchuk, Oleksii Kuchaiev, Vitaly Lavrukhin, Ryan Leary, Jason Li, Huyen Nguyen, Jonathan M. Cohen

Figure 1 for Stochastic Gradient Methods with Layer-wise Adaptive Moments for Training of Deep Networks
Figure 2 for Stochastic Gradient Methods with Layer-wise Adaptive Moments for Training of Deep Networks
Figure 3 for Stochastic Gradient Methods with Layer-wise Adaptive Moments for Training of Deep Networks
Figure 4 for Stochastic Gradient Methods with Layer-wise Adaptive Moments for Training of Deep Networks
* Submitted to NeurIPS 2019 
  Access Paper or Ask Questions

Jasper: An End-to-End Convolutional Neural Acoustic Model


Apr 05, 2019
Jason Li, Vitaly Lavrukhin, Boris Ginsburg, Ryan Leary, Oleksii Kuchaiev, Jonathan M. Cohen, Huyen Nguyen, Ravi Teja Gadde

Figure 1 for Jasper: An End-to-End Convolutional Neural Acoustic Model
Figure 2 for Jasper: An End-to-End Convolutional Neural Acoustic Model
Figure 3 for Jasper: An End-to-End Convolutional Neural Acoustic Model
Figure 4 for Jasper: An End-to-End Convolutional Neural Acoustic Model
* Submitted to Interspeech 2019 
  Access Paper or Ask Questions

Training Neural Speech Recognition Systems with Synthetic Speech Augmentation


Nov 02, 2018
Jason Li, Ravi Gadde, Boris Ginsburg, Vitaly Lavrukhin

Figure 1 for Training Neural Speech Recognition Systems with Synthetic Speech Augmentation
Figure 2 for Training Neural Speech Recognition Systems with Synthetic Speech Augmentation
Figure 3 for Training Neural Speech Recognition Systems with Synthetic Speech Augmentation
Figure 4 for Training Neural Speech Recognition Systems with Synthetic Speech Augmentation
* Pre-print. Work in progress, 5 pages, 1 figure 
  Access Paper or Ask Questions

OpenSeq2Seq: extensible toolkit for distributed and mixed precision training of sequence-to-sequence models


May 25, 2018
Oleksii Kuchaiev, Boris Ginsburg, Igor Gitman, Vitaly Lavrukhin, Carl Case, Paulius Micikevicius

Figure 1 for OpenSeq2Seq: extensible toolkit for distributed and mixed precision training of sequence-to-sequence models
Figure 2 for OpenSeq2Seq: extensible toolkit for distributed and mixed precision training of sequence-to-sequence models
Figure 3 for OpenSeq2Seq: extensible toolkit for distributed and mixed precision training of sequence-to-sequence models
Figure 4 for OpenSeq2Seq: extensible toolkit for distributed and mixed precision training of sequence-to-sequence models
* to be presented at Workshop for Natural Language Processing Open Source Software (NLP-OSS), co-located with ACL2018 
  Access Paper or Ask Questions

Factorization tricks for LSTM networks


Feb 24, 2018
Oleksii Kuchaiev, Boris Ginsburg

Figure 1 for Factorization tricks for LSTM networks
Figure 2 for Factorization tricks for LSTM networks
Figure 3 for Factorization tricks for LSTM networks
* accepted to ICLR 2017 Workshop 
  Access Paper or Ask Questions

Mixed Precision Training


Feb 15, 2018
Paulius Micikevicius, Sharan Narang, Jonah Alben, Gregory Diamos, Erich Elsen, David Garcia, Boris Ginsburg, Michael Houston, Oleksii Kuchaiev, Ganesh Venkatesh, Hao Wu

Figure 1 for Mixed Precision Training
Figure 2 for Mixed Precision Training
Figure 3 for Mixed Precision Training
Figure 4 for Mixed Precision Training
* Published as a conference paper at ICLR 2018 
  Access Paper or Ask Questions

Training Deep AutoEncoders for Collaborative Filtering


Oct 10, 2017
Oleksii Kuchaiev, Boris Ginsburg

Figure 1 for Training Deep AutoEncoders for Collaborative Filtering
Figure 2 for Training Deep AutoEncoders for Collaborative Filtering
Figure 3 for Training Deep AutoEncoders for Collaborative Filtering
Figure 4 for Training Deep AutoEncoders for Collaborative Filtering
* 5 pages, 6 figures 
  Access Paper or Ask Questions

Comparison of Batch Normalization and Weight Normalization Algorithms for the Large-scale Image Classification


Oct 07, 2017
Igor Gitman, Boris Ginsburg

Figure 1 for Comparison of Batch Normalization and Weight Normalization Algorithms for the Large-scale Image Classification
Figure 2 for Comparison of Batch Normalization and Weight Normalization Algorithms for the Large-scale Image Classification
Figure 3 for Comparison of Batch Normalization and Weight Normalization Algorithms for the Large-scale Image Classification
Figure 4 for Comparison of Batch Normalization and Weight Normalization Algorithms for the Large-scale Image Classification

  Access Paper or Ask Questions

Large Batch Training of Convolutional Networks


Sep 13, 2017
Yang You, Igor Gitman, Boris Ginsburg

Figure 1 for Large Batch Training of Convolutional Networks
Figure 2 for Large Batch Training of Convolutional Networks
Figure 3 for Large Batch Training of Convolutional Networks
Figure 4 for Large Batch Training of Convolutional Networks

  Access Paper or Ask Questions

SEBOOST - Boosting Stochastic Learning Using Subspace Optimization Techniques


Sep 02, 2016
Elad Richardson, Rom Herskovitz, Boris Ginsburg, Michael Zibulevsky

Figure 1 for SEBOOST - Boosting Stochastic Learning Using Subspace Optimization Techniques
Figure 2 for SEBOOST - Boosting Stochastic Learning Using Subspace Optimization Techniques
Figure 3 for SEBOOST - Boosting Stochastic Learning Using Subspace Optimization Techniques
Figure 4 for SEBOOST - Boosting Stochastic Learning Using Subspace Optimization Techniques

  Access Paper or Ask Questions