Prominent methods (e.g., Tacotron 2)usuallyfirst generate mel-spectrogram from text, and then synthesize speech from themel-spectrogram using vocoder such as WaveNet. Compared with traditionalconcatenative and statistical parametric approaches, neural network based end-to-end models suffer from slow inference speed, and the synthesized speech isusually not robust (i.e., some words are skipped or repeated) and lack of con-trollability (voice speed or prosody control).

Neural network based end-to-end text to speech (TTS) has significantly improved the quality of synthesized speech. Prominent methods (e.g., Tacotron 2) usually first generate mel-spectrogram from text, and then synthesize speech from the mel-spectrogram using vocoder such as WaveNet. Compared with traditional concatenative and statistical parametric approaches, neural network based end-to-end models suffer from slow inference speed, and the synthesized speech is usually not robust (i.e., some words are skipped or repeated) and lack of controllability (voice speed or prosody control). In this work, we propose a novel feed-forward network based on Transformer to generate mel-spectrogram in parallel for TTS. Specifically, we extract attention alignments from an encoder-decoder based teacher model for phoneme duration prediction, which is used by a length regulator to expand the source phoneme sequence to match the length of the target mel-spectrogram sequence for parallel mel-spectrogram generation. Experiments on the LJSpeech dataset show that our parallel model matches autoregressive models in terms of speech quality, nearly eliminates the problem of word skipping and repeating in particularly hard cases, and can adjust voice speed smoothly. Most importantly, compared with autoregressive Transformer TTS, our model speeds up mel-spectrogram generation by 270x and the end-to-end speech synthesis by 38x. Therefore, we call our model FastSpeech.

Thanks all the reviewers for the detailed and thoughtful comments. HMM-based works [1, 2, 3], all of which proposed methods to estimate alignments from unsegmented data. We've not thoroughly explored to improve the duration predictor and simply follow the same We design the grouped 1x1 convolutions to be able to mix channels. For example, to generate a speech of 5.8 Therefore, adopting parallel TTS models significantly improves the sampling speed of end-to-end systems. In Section 5.3, we showed that varying temperature can change We will add a reference about Viterbi training.

Prominent methods (e.g., Tacotron 2) usually first generate mel-spectrogram from text, and then synthesize speech from the mel-spectrogram using vocoder such as WaveNet. Compared with traditional concatenative and statistical parametric approaches, neural network based end-to-end models suffer from slow inference speed, and the synthesized speech is usually not robust (i.e., some words are skipped or repeated) and lack of con-trollability (voice speed or prosody control).

Originally: Although phoneme duration prediction is widely adopted in conventional TTS systems, jointly training it in a neural TTS model is new. This paper is one of the first works on non-autoregressive text-to-spectrogram modeling. Quality: This paper seems sound overall, expected for a few issues in the comments below. Some of these issues must be addressed before acceptance. Clarity: A well written paper. Significance: The advantages over its autoregressive counterparts are significant, especially for industrial use.

The paper proposes a novel non-autoregressive parallelisation approach for mel-spectrogram intermediate representation TTS. The reviewers concur that the paper incorporates two novel explicit components to tts systems - length and duration modules and that the results on Speedup at inference and high-quality audio generations are relevant.

Neural network based end-to-end text to speech (TTS) has significantly improved the quality of synthesized speech. Prominent methods (e.g., Tacotron 2) usually first generate mel-spectrogram from text, and then synthesize speech from the mel-spectrogram using vocoder such as WaveNet. Compared with traditional concatenative and statistical parametric approaches, neural network based end-to-end models suffer from slow inference speed, and the synthesized speech is usually not robust (i.e., some words are skipped or repeated) and lack of controllability (voice speed or prosody control). In this work, we propose a novel feed-forward network based on Transformer to generate mel-spectrogram in parallel for TTS. Specifically, we extract attention alignments from an encoder-decoder based teacher model for phoneme duration prediction, which is used by a length regulator to expand the source phoneme sequence to match the length of the target mel-spectrogram sequence for parallel mel-spectrogram generation.

Recently, denoising diffusion probabilistic models and generative score matching have shown high potential in modelling complex data distributions while stochastic calculus has provided a unified point of view on these techniques allowing for flexible inference schemes. In this paper we introduce Grad-TTS, a novel text-to-speech model with score-based decoder producing mel-spectrograms by gradually transforming noise predicted by encoder and aligned with text input by means of Monotonic Alignment Search. The framework of stochastic differential equations helps us to generalize conventional diffusion probabilistic models to the case of reconstructing data from noise with different parameters and allows to make this reconstruction flexible by explicitly controlling trade-off between sound quality and inference speed. Subjective human evaluation shows that Grad-TTS is competitive with state-of-the-art text-to-speech approaches in terms of Mean Opinion Score. We will make the code publicly available shortly.

Neural network based end-to-end text to speech (TTS) has significantly improved the quality of synthesized speech. Prominent methods (e.g., Tacotron 2) usually first generate mel-spectrogram from text, and then synthesize speech from the mel-spectrogram using vocoder such as WaveNet. Compared with traditional concatenative and statistical parametric approaches, neural network based end-to-end models suffer from slow inference speed, and the synthesized speech is usually not robust (i.e., some words are skipped or repeated) and lack of controllability (voice speed or prosody control). In this work, we propose a novel feed-forward network based on Transformer to generate mel-spectrogram in parallel for TTS. Specifically, we extract attention alignments from an encoder-decoder based teacher model for phoneme duration prediction, which is used by a length regulator to expand the source phoneme sequence to match the length of the target mel-spectrogram sequence for parallel mel-spectrogram generation. Experiments on the LJSpeech dataset show that our parallel model matches autoregressive models in terms of speech quality, nearly eliminates the problem of word skipping and repeating in particularly hard cases, and can adjust voice speed smoothly.

Text to speech (TTS) has attracted a lot of attention recently due to advancements in deep learning. Neural network-based TTS models (such as Tacotron 2, DeepVoice 3 and Transformer TTS) have outperformed conventional concatenative and statistical parametric approaches in terms of speech quality. Neural network-based TTS models usually first generate a mel-scale spectrogram (or mel-spectrogram) autoregressively from text input and then synthesize speech from the mel-spectrogram using a vocoder. A spectrogram is a visual representation of frequencies measured over time.) To address the above problems, researchers from Microsoft and Zhejiang University propose FastSpeech, a novel feed-forward network that generates mel-spectrograms with fast generation speed, robustness, controllability, and high quality.