AI People Wiki

Builders & open source

Jonathan Shen

First author of the Tacotron 2 text-to-speech paper, written as a software engineer at Google

Jonathan Shen is first author of Natural TTS Synthesis by Conditioning WaveNet on Mel Spectrogram Predictions, the Tacotron 2 paper, which pairs a recurrent sequence-to-sequence network that maps character embeddings to mel-scale spectrograms with a modified WaveNet vocoder that synthesizes waveforms from them. Shen co-wrote the December 2017 Google Research blog post introducing Tacotron 2, posted with Ruoming Pang as software engineers on behalf of the Google Brain and Machine Perception teams.12

At a glance

Current role
None confirmed in the sources
Work listed
1
Sources cited
7
Profile reviewed

Speech synthesis research

Shen was first author of the December 2017 Tacotron 2 paper, which combines a sequence-to-sequence network predicting mel spectrograms from characters with a modified WaveNet vocoder. The Google Research blog post introducing the system, posted by Shen and Ruoming Pang as software engineers for the Google Brain and Machine Perception teams, also described open problems such as pronouncing complex words and controlling the style of generated speech.12

Shen co-authored later Google text-to-speech papers, including a 2018 system that transfers knowledge from speaker verification to multispeaker speech synthesis and a 2018 paper on hierarchical generative modeling for controllable synthesis. In 2020 Shen was first author of Non-Attentive Tacotron, which replaces the attention mechanism of Tacotron 2 with explicit duration prediction, including an unsupervised duration model.345

Open-source work

Shen was first author of the 2019 paper describing Lingvo, a TensorFlow framework for collaborative deep learning research with a focus on sequence-to-sequence models. Lingvo's code is published in the tensorflow/lingvo repository under the Apache-2.0 license.67

Work and contributions

  • Tacotron 2 Published research

    First author of Natural TTS Synthesis by Conditioning WaveNet on Mel Spectrogram Predictions (2017) and co-author of the Google Research blog post presenting Tacotron 2.12

Timeline

  1. Posted the Tacotron 2 paper as first author.1
  2. Co-wrote the Google Research blog post introducing Tacotron 2 with Ruoming Pang.2
  3. Co-authored "Transfer Learning from Speaker Verification to Multispeaker Text-To-Speech Synthesis".3
  4. Co-authored "Hierarchical Generative Modeling for Controllable Speech Synthesis".4
  5. Posted the Lingvo sequence-to-sequence framework paper as first author.6
  6. Posted "Non-Attentive Tacotron" as first author.5

Documented roles

The sources do not confirm a current role; these are the roles they document.

Companies in their history

Suggest a correction or removal

Profiles cover public professional work only, and each statement cites a source. If something is out of date, missing or wrong, tell us and link a source. People can also ask for their profile to be corrected or removed.

References

7 sources · profile reviewed Oct 21, 2026 · data updated Oct 21, 2026

  1. Natural TTS Synthesis by Conditioning WaveNet on Mel Spectrogram Predictions (arXiv 1712.05884)arXiv · Paper · published Dec 16, 2017 · accessed Oct 21, 2026
  2. Tacotron 2: Generating Human-like Speech from Text (Google Research blog)Google Research · Announcement · published Dec 19, 2017 · accessed Oct 21, 2026
  3. Transfer Learning from Speaker Verification to Multispeaker Text-To-Speech Synthesis (arXiv 1806.04558)arXiv · Paper · published Jun 12, 2018 · accessed Oct 21, 2026
  4. Hierarchical Generative Modeling for Controllable Speech Synthesis (arXiv 1810.07217)arXiv · Paper · published Oct 16, 2018 · accessed Oct 21, 2026
  5. Non-Attentive Tacotron: Robust and Controllable Neural TTS Synthesis Including Unsupervised Duration Modeling (arXiv 2010.04301)arXiv · Paper · published Oct 8, 2020 · accessed Oct 21, 2026
  6. Lingvo: a Modular and Scalable Framework for Sequence-to-Sequence Modeling (arXiv 1902.08295)arXiv · Paper · published Feb 21, 2019 · accessed Oct 21, 2026
  7. tensorflow/lingvo repository (Apache-2.0 license)TensorFlow (GitHub repository) · Repository · accessed Oct 21, 2026