Tacotron: A Fully End-to-End Text-To-Speech Synthesis Model

By Yuxuan Wang, R. Skerry-Ryan, Daisy Stanton, Yonghui Wu, Ron J. Weiss, Navdeep Jaitly, Zongheng Yang, Ying Xiao, Z. Chen, Samy Bengio and 4 more. 2017

255 citations recorded by the source corpus.

Subfields: Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Natural Language Processing

Citations in the corpus

Waveglow: A Flow-based Generative Network for Speech Synthesis (2018, background): “The ﬁrst step transforms the text into time-aligned features, such as a mel-spectrogram [4, 5], or F0 frequencies and other linguistic features [2, 6].”
Efficient Neural Audio Synthesis (2018, background): “Convolutional sequence models (Kalchbrenner et al., 2016) achieve excellent performance in speech synthesis (Wang et al., 2017), yet their architecture tends to be deep and narrow requiring a long chain of layers to be executed for each sample.”
Neural Speech Synthesis with Transformer Network (2018, background): “Recently, with the rapid development of neural networks, end-to-end generative text-to-speech models, such as Tacotron (Wang et al. 2017) and Tacotron2 (Shen et al. 2017), are proposed to simplify traditional speech synthesis pipeline by replacing the production of these linguistic and acoustic…”
Data Augmentation Generative Adversarial Networks (2017, background): “…machine translation (Wu et al., 2016), natural language processing (Gu et al., 2015), speech recognition (Hinton et al., 2012a), and synthesis (Wang et al., 2017), learning from human play (Clark & Storkey, 2015) and reinforcement learning (Mnih et al., 2015; Silver et al., 2016; Foerster et…”
Neural Voice Cloning with a Few Samples (2018, background): “neralize to unseen texts. Equal contribution 2Cloned audio samples can be found in https://audiodemos.github.io 3A single speaker model can require ˘20 hours of training data [e.g.,Arik et al. ,2017a Wang et al. 2017], while a multi-speaker model for 108 speakers [Arik et al.,2017b] requires about ˘”
Fitting New Speakers Based on a Short Untranscribed Sample (2018, neutral): “The recent neural TTS systems include the Deep Voice systems DV1 (Arik et al., 2017b), DV2 (Arik et al., 2017a) and DV3 (Ping et al., 2018), WaveNet (Oord et al., 2016) and Parallel WaveNet (van den Oord et al., 2017), Char2Wav (Sotelo et al., 2017), Tacotron (Wang et al., 2017) and Tacotron2 (Shen ”
Fast Spectrogram Inversion Using Multi-Head Convolutional Neural Networks (2018, methodology): “, [10], [11], synthesize audio by applying autoregression at the rate of spectrogram timeframes (100 s of samples), and still does not yield a noticeable decrease in audio quality.”
Flowtron: an Autoregressive Flow-based Generative Network for Text-to-Speech Synthesis (2020, methodology): “, 2017) require adding a critical Prenet layer to help with convergence and improve generalization (Wang et al., 2017).”
Voice Synthesis for in-the-Wild Speakers via a Phonological Loop (2017, methodology): “Trained on a large training set, the Tacotron system is able to read raw text (characters and not phonemes).”
Listening while speaking: Speech chain by deep learning (2017, methodology): “In this work, our core architecture is based on Tacotron [23] with several structural modiﬁcations.”
Statistical parametric speech synthesis using generative adversarial networks under a multi-task learning framework (2017, methodology): “In our future work, we will focus on improving the performance of the generator in GAN and try to use GAN in end-to-end speech synthesis [33, 34, 35, 36, 37].”
Generative adversarial network-based approach to signal reconstruction from magnitude spectrogram (2018, methodology): “Furthermore, a magnitude spectrogram representation was recently found to be reasonable and effective for use with speech synthesis systems [4], [5].”
Sequence-to-sequence Modelling of F0 for Speech Emotion Conversion (2019, methodology): “Sequence-to-sequence models have recently been applied with success for text-to-speech synthesis [21, 22], approaching WaveNet [23] in terms of quality at lower computational cost and latency.”
Controlling Emotion Strength with Relative Attribute for End-to-End Speech Synthesis (2019, unknown): “And the strength is normalized to [0, 1] for ease-of-use.”
EMPHASIS: An Emotional Phoneme-based Acoustic Model for Speech Synthesis System (2018, methodology): “D convolution banks, highway layers and bi-directional GRU layers (CBHG) [9], [13].”
Efficiently Trainable Text-to-Speech System Based on Deep Convolutional Networks with Guided Attention (2017, background): “Indeed, some people tried to implement open clones of Tacotron [3, 4, 5, 6], but they are struggling to reproduce the speech of satisfactory quality as clear as the original work.”
Augmenting Image Classifiers Using Data Augmentation Generative Adversarial Networks (2018, background): “They have been demonstrated in variety of domains [12] spanning from image classification [25, 17, 16, 18, 21], machine translation [44], natural language processing [12], speech recognition [19], and synthesis [42], learning from human play [6] and reinforcement learning [27, 35, 10, 40, 13] among ”