Baidu USA LLC

Sanjeev Satheesh

Adam Coates

Bryan Catanzaro

Ryan Prenger

Zhenyao Zhu

Shubhabrata Sengupta

Awni Hannun

Anuroop Sriram

Erich Elsen

Christopher Fougner

Heewoo Jun

Xu Han

Gregory Diamos

Jun Zhan

Jesse Engel

Chong Wang

Hairong Liu

Andrew Ng

Mike Chrzanowski

Dario Amodei

Carl Case

Jared Casper

Yashesh Gaur

Jingdong Chen

Dani Yogatama

Eric Dean Battenberg

Hee Woo Jun

Vinay Rao

Markus Kliegl

Rewon Child

Ajay Kannan

Jiaji Huang

Atul Kumar

David Seetapun

Erich Eisen

Sengupta Shubhabrata

Described herein are systems and methods for generating natural language sentences with Sequence-to-sequence (Seq2Seq) models with attention. The Seq2Seq models may be implemented in applications, such as machine translation, image captioning, and speech recognition. Performance has further been improved by leveraging unlabeled data, often in the form of a language models. Disclosed herein are 'Cold Fusion' architecture embodiments that leverage a pre-trained language model during training. The Seq2Seq models with Cold Fusion embodiments are able to better utilize language information enjoying faster convergence, better generalization, and almost complete transfer to a new domain while using less labeled training data.

Cold fusing sequence-to-sequence models with language models

Presented herein are embodiments of state-of-the-art speech recognition systems developed using end-to-end deep learning. In embodiments, the model architecture is significantly simpler than traditional speech systems, which rely on laboriously engineered processing pipelines; these traditional systems also tend to perform poorly when used in noisy environments. In contrast, embodiments of the system do not need hand-designed components to model background noise, reverberation, or speaker variation, but instead directly learn a function that is robust to such effects. Neither a phoneme dictionary, nor even the concept of a 'phoneme,' is needed. Embodiments include a well-optimized recurrent neural network (RNN) training system that can use multiple GPUs, as well as a set of novel data synthesis techniques that allows for a large amount of varied data for training to be efficiently obtained. Embodiments of the system can also handle challenging noisy environments better than widely used, state-of-the-art commercial speech systems.

Deep learning models for speech recognition

Described herein are systems and methods for a general, scalable, end-to-end framework that uses a generative adversarial network (GAN) objective to enable robust speech recognition. Encoders trained with the proposed approach enjoy improved invariance by learning to map noisy audio to the same embedding space as that of clean audio. Embodiments of a Wasserstein GAN framework increase the robustness of seq-to-seq models in a scalable, end-to-end fashion. In one or more embodiments, an encoder component is treated as the generator of GAN and is trained to produce indistinguishable embeddings between labeled and unlabeled audio samples. This new robust training approach can learn to induce robustness without alignment or complicated inference pipeline and even where augmentation of audio data is not possible.

Systems and methods for robust speech recognition using generative adversarial networks

Described herein are systems and methods to identify and address sources of bias in an end-to-end speech model. In one or more embodiments, the end-to-end model may be a recurrent neural network with two 2D-convolutional input layers, followed by multiple bidirectional recurrent layers and one fully connected layer before a softmax layer. In one or more embodiments, the network is trained end-to-end using the CTC loss function to directly predict sequences of characters from log spectrograms of audio. With optimized recurrent layers and training together with alignment information, some unwanted bias induced by using purely forward only recurrences may be removed in a deployed model.

Systems and methods for principled bias reduction in production speech models

Presented herein are embodiments of state-of-the-art speech recognition systems developed using end-to-end deep learning. In embodiments, the model architecture is significantly simpler than traditional speech systems, which rely on laboriously engineered processing pipelines; these traditional systems also tend to perform poorly when used in noisy environments. In contrast, embodiments of the system do not need hand-designed components to model background noise, reverberation, or speaker variation, but instead directly learn a function that is robust to such effects. A phoneme dictionary, nor even the concept of a 'phoneme,' is needed. Embodiments include a well-optimized recurrent neural network (RNN) training system that can use multiple GPUs, as well as a set of novel data synthesis techniques that allows for a large amount of varied data for training to be efficiently obtained. Embodiments of the system can also handle challenging noisy environments better than widely used, state-of-the-art commercial speech systems.

Systems and methods for speech transcription

Described herein are systems and methods for automatic unit selection and target decomposition for sequence labelling. Embodiments include a new loss function called Gram-Connectionist Temporal Classification (CTC) loss that extend the popular CTC loss function criterion to alleviate prior limitations. While preserving the advantages of CTC, Gram-CTC automatically learns the best set of basic units (grams), as well as the most suitable decomposition of target sequences. Unlike CTC, embodiments of Gram-CTC allow a model to output variable number of characters at each time step, which enables the model to capture longer term dependency and improves the computational efficiency. It is also demonstrated that embodiments of Gram-CTC improve CTC in terms of both performance and efficiency on the large vocabulary speech recognition task at multiple scales of data, and that systems that employ an embodiment of Gram-CTC can outperform the state-of-the-art on a standard speech benchmark.

Systems and methods for automatic unit selection and target decomposition for sequence labelling

Embodiments of end-to-end deep learning systems and methods are disclosed to recognize speech of vastly different languages, such as English or Mandarin Chinese. In embodiments, the entire pipelines of hand-engineered components are replaced with neural networks, and the end-to-end learning allows handling a diverse variety of speech including noisy environments, accents, and different languages. Using a trained embodiment and an embodiment of a batch dispatch technique with GPUs in a data center, an end-to-end deep learning system can be inexpensively deployed in an online setting, delivering low latency when serving users at scale.

End-to-end speech recognition

Deployed end-to-end speech recognition

Growing community of inventors

Sunnyvale, CA, United States of America

Sanjeev Satheesh