Baidu USA LLC

Christopher Fougner

Bryan Catanzaro

Ryan Prenger

Sanjeev Satheesh

Zhenyao Zhu

Xu Han

Jun Zhan

Adam Coates

Jesse Engel

Erich Elsen

Chong Wang

Awni Hannun

Shubhabrata Sengupta

Mike Chrzanowski

Markus Kliegl

Dario Amodei

Rewon Child

Jingdong Chen

Dani Yogatama

Sercan Omer Arik

Andrew Gibiansky

Eric Dean Battenberg

Anuroop Sriram

Hairong Liu

Joel Hestness

Vinay Rao

Heewoo Jun

Ajay Kannan

Yashesh Gaur

Jiaji Huang

Atul Kumar

David Seetapun

Disclosed are systems and methods that implement efficient engines for computation-intensive tasks such as neural network deployment. Various embodiments of the invention provide for high-throughput batching that increases throughput of streaming data in high-traffic applications, such as real-time speech transcription. In embodiments, throughput is increased by dynamically assembling into batches and processing together user requests that randomly arrive at unknown timing such that not all the data is present at once at the time of batching. Some embodiments allow for performing steaming classification using pre-processing. The gains in performance allow for more efficient use of a compute engine and drastically reduce the cost of deploying large neural networks at scale, while meeting strict application requirements and adding relatively little computational latency so as to maintain a satisfactory application experience.

Systems and methods for efficient neural network deployments

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

Described herein are systems and methods for creating and using Convolutional Recurrent Neural Networks (CRNNs) for small-footprint keyword spotting (KWS) systems. Inspired by the large-scale state-of-the-art speech recognition systems, in embodiments, the strengths of convolutional layers to utilize the structure in the data in time and frequency domains are combined with recurrent layers to utilize context for the entire processed frame. The effect of architecture parameters were examined to determine preferred model embodiments given the performance versus model size tradeoff. Various training strategies are provided to improve performance. In embodiments, using only &#x2dc;230 k parameters and yielding acceptably low latency, a CRNN model embodiment demonstrated high accuracy and robust performance in a wide range of environments.

Convolutional recurrent neural networks for small-footprint keyword spotting

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

Palo Alto, CA, United States of America

Christopher Fougner