HackerPapers

arXiv:1512.02595cs.CL

Deep Speech 2: End-to-End Speech Recognition in English and Mandarin

Dario Amodei, Rishita Anubhai, Eric Battenberg, Carl Case, Jared Casper, Bryan Catanzaro, Jingdong Chen, Mike Chrzanowski, Adam Coates, Greg Diamos, Erich Elsen, Jesse Engel, Linxi Fan, Christopher Fougner, Tony Han, Awni Hannun, Billy Jun, Patrick LeGresley, Libby Lin, Sharan Narang, Andrew Ng, Sherjil Ozair, Ryan Prenger, Jonathan Raiman, Sanjeev Satheesh, David Seetapun, Shubho Sengupta, Yi Wang, Zhiqian Wang, Chong Wang, Bo Xiao, Dani Yogatama, Jun Zhan, Zhenyao Zhu

Abstract

We show that an end-to-end deep learning approach can be used to recognize either English or Mandarin Chinese speech--two vastly different languages. Because it replaces entire pipelines of hand-engineered components with neural networks, end-to-end learning allows us to handle a diverse variety of speech including noisy environments, accents and different languages. Key to our approach is our application of HPC techniques, resulting in a 7x speedup over our previous system. Because of this efficiency, experiments that previously took weeks now run in days. This enables us to iterate more quickly to identify superior architectures and algorithms. As a result, in several cases, our system is competitive with the transcription of human workers when benchmarked on standard datasets. Finally, using a technique called Batch Dispatch with GPUs in the data center, we show that our system can be inexpensively deployed in an online setting, delivering low latency when serving users at scale.

arXiv comments
Journal reference
DOI
Subjects
cs.CLComputation and Language

On Hacker News

Linked by 10 identifiable commenters in 10 comments across 9 threads, from Jul 2016 to Jul 2026. Found in 1 HN submission from Dec 2015 to Dec 2015. Open submission discussions.

Comments linking this paper, per year.

Discussed in comments9 threads

  1. 30papers.com – Ilya's 30 essential ML papers, in a beginner friendly format 2 comments

    • https://colah.github.io/posts/2015-08-Understanding-LSTMs/ https://papers.nips.cc/paper/2012/hash/c399862d3b9d6b76c8436... https://papers.nips.cc/paper_files/paper/2012/file/c399862d3... https://arxiv.org/pdf/1512.03385 https://arxiv.org/pdf/1511.07122…

      gooobView on HNParent context

    • No need stupid moving texts. CS231n: Convolutional Neural Networks for Visual Recognition - https://cs231n.github.io/ The Unreasonable Effectiveness of Recurrent Neural Networks - https://karpathy.github.io/2015/05/21/rnn-effectiveness/ Understanding LSTM Networks…

      cute_boiView on HNParent context

  2. Speech and Language Processing (3rd ed. draft) 1 comment

    • The acoustic model of a speech recognizer used to be a GMM, which mapped a pre-processed acoustic signal vector (generally MFCCs-Mel-Frequency Cepstral Coefficients) to an HMM state. Now those layers are neural nets, so acoustic pre-processing, GMM, and HMM are all subsumed by…

      brandonbView on HNParent context

  3. Ilya Sutskever: “If you learn all of these, you’ll know 90% of what matters” 1 comment

    • a backup just in case https://nlp.seas.harvard.edu/annotated-transformer/ https://scottaaronson.blog/?p=762 https://karpathy.github.io/2015/05/21/rnn-effectiveness/ https://colah.github.io/posts/2015-08-Understanding-LSTMs/ https://arxiv.org/pdf/1409.2329.pdf…

      is_trueView on HNParent context

  4. Has Google Translate been fixed yet? 1 comment
  5. RWKV: Reinventing RNNs for the Transformer Era 1 comment
  6. Common Voice: A Massively-Multilingual Speech Corpus 1 comment

    • > M-AILABS data contains 9 language varieties with a modified BSD 3-Clause License, however there is no community-driven aspect. There is certainly an aspect, as M-AILABS sources their data mainly from the librivox project, which is community-driven. To give ballpark numbers on…

      est31View on HNParent context

  7. DeepSpeech 0.6 1 comment

    • But these results are about in line with other CTC systems. The original Baidu paper [1] had 7.89% with original Deep Speech and 5.33% with "Deep Speech 2" where they threw more complicated networks at it and added another 11K hours of external speech traingin data. [1]…

      gokView on HNParent context

  8. Initial Release of Mozilla’s Open Source Speech Recognition Model and Voice Data 1 comment

    • In the Baidu Deep Speech 2 paper, the Baidu implementation is able to get 5.33%, and a human 5.83%. https://arxiv.org/pdf/1512.02595v1.pdf

      TD-LinuxView on HNParent context

  9. Say_what: Using speech-to-text to fully check out during conference calls 1 comment

    • * I should clarify as well, for your first question. State of the art is generally either IBM's 2016 English Conversational Telephone Speech Recognition System [0], or Baidu's DeepSpeech2 [1]. IBM's is a little more complex, but I'd actually argue Baidu is more promising due to…

      skoocdaView on HNParent context

Submitted to Hacker News

Connected through shared HN threads, not inferred similarity or a recommendation.