# Predicting phoneme-level prosody latents using AR and flow-based Prior Networks for expressive speech synthesis

arXiv: https://arxiv.org/abs/2211.01327 · HTML: https://mlsanity.com/paper/2211.01327/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 27 of its 32 references that are in the map.
- First posted on arXiv 2 Nov 2022.

## Builds on (most cited first, 27 of 27)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Glow: Generative Flow with Invertible 1x1 Convolutions](https://mlsanity.com/paper/1807.03039/) (2018)
- [Natural TTS Synthesis by Conditioning WaveNet on Mel Spectrogram Predictions](https://mlsanity.com/paper/1712.05884/) (2017)
- [Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech](https://mlsanity.com/paper/2106.06103/) (2021)
- [Tacotron: Towards End-to-End Speech Synthesis](https://mlsanity.com/paper/1703.10135/) (2017)
- [A Recurrent Latent Variable Model for Sequential Data](https://mlsanity.com/paper/1506.02216/) (2015)
- [Glow-TTS: A Generative Flow for Text-to-Speech via Monotonic Alignment Search](https://mlsanity.com/paper/2005.11129/) (2020)
- [Flow++: Improving Flow-Based Generative Models with Variational Dequantization and Architecture Design](https://mlsanity.com/paper/1902.00275/) (2019)
- [Style Tokens: Unsupervised Style Modeling, Control and Transfer in End-to-End Speech Synthesis](https://mlsanity.com/paper/1803.09017/) (2018)
- [Dynamical Variational Autoencoders: A Comprehensive Review](https://mlsanity.com/paper/2008.12595/) (2020)
- [Emformer: Efficient Memory Transformer Based Acoustic Model For Low Latency Streaming Speech Recognition](https://mlsanity.com/paper/2010.10759/) (2020)
- [LPCNet: Improving Neural Speech Synthesis Through Linear Prediction](https://mlsanity.com/paper/1810.11846/) (2018)
- [Non-Attentive Tacotron: Robust and Controllable Neural TTS Synthesis Including Unsupervised Duration Modeling](https://mlsanity.com/paper/2010.04301/) (2020)
- [Learning Stochastic Recurrent Networks](https://mlsanity.com/paper/1411.7610/) (2014)
- [Robust and fine-grained prosody control of end-to-end speech synthesis](https://mlsanity.com/paper/1811.02122/) (2018)
- [Fully-hierarchical fine-grained prosody modeling for interpretable speech synthesis](https://mlsanity.com/paper/2002.03785/) (2020)
- [VFlow: More Expressive Generative Flows with Variational Data Augmentation](https://mlsanity.com/paper/2002.09741/) (2020)
- [Generating diverse and natural text-to-speech samples using a quantized fine-grained VAE and auto-regressive prosody prior](https://mlsanity.com/paper/2002.03788/) (2020)
- [Predicting Expressive Speaking Style From Text In End-To-End Speech Synthesis](https://mlsanity.com/paper/1808.01410/) (2018)
- [ProsoSpeech: Enhancing Prosody With Quantized Vector Pre-training in Text-to-Speech](https://mlsanity.com/paper/2202.07816/) (2022)
- [High Quality Streaming Speech Synthesis with Low, Sentence-Length-Independent Latency](https://mlsanity.com/paper/2111.09052/) (2021)
- [Hierarchical Prosody Modeling for Non-Autoregressive Speech Synthesis](https://mlsanity.com/paper/2011.06465/) (2020)
- [Prosody Transfer in Neural Text to Speech Using Global Pitch and Loudness Features](https://mlsanity.com/paper/1911.09645/) (2019)
- [Prosodic Clustering for Phoneme-level Prosody Control in End-to-End Speech Synthesis](https://mlsanity.com/paper/2111.10177/) (2021)
- [Unsupervised word-level prosody tagging for controllable speech synthesis](https://mlsanity.com/paper/2202.07200/) (2022)
- [Expressive, Variable, and Controllable Duration Modelling in TTS](https://mlsanity.com/paper/2206.14165/) (2022)
- [Word-Level Style Control for Expressive, Non-attentive Speech Synthesis](https://mlsanity.com/paper/2111.10173/) (2021)

## Abstract (arXiv)

A large part of the expressive speech synthesis literature focuses on learning prosodic representations of the speech signal which are then modeled by a prior distribution during inference. In this paper, we compare different prior architectures at the task of predicting phoneme level prosodic representations extracted with an unsupervised FVAE model. We use both subjective and objective metrics to show that normalizing flow based prior networks can result in more expressive speech at the cost of a slight drop in quality. Furthermore, we show that the synthesized speech has higher variability, for a given text, due to the nature of normalizing flows. We also propose a Dynamical VAE model, that can generate higher quality speech although with decreased expressiveness and variability compared to the flow based models.