# Phone-Level Prosody Modelling with GMM-Based MDN for Diverse and Controllable Speech Synthesis

arXiv: https://arxiv.org/abs/2105.13086 · HTML: https://mlsanity.com/paper/2105.13086/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 27 May 2021.

## Built on by (most cited first, 10 of 10)

- [EmoDiff: Intensity Controllable Emotional Text-to-Speech with Soft-Label Guidance](https://mlsanity.com/paper/2211.09496/) (2022)
- [Autoregressive Diffusion Transformer for Text-to-Speech Synthesis](https://mlsanity.com/paper/2406.05551/) (2024)
- [PromptTTS++: Controlling Speaker Identity in Prompt-Based Text-to-Speech Using Natural Language Descriptions](https://mlsanity.com/paper/2309.08140/) (2023)
- [ControlSpeech: Towards Simultaneous and Independent Zero-shot Speaker Cloning and Zero-shot Language Style Control](https://mlsanity.com/paper/2406.01205/) (2024)
- [DiffProsody: Diffusion-based Latent Prosody Generation for Expressive Speech Synthesis with Prosody Conditional Adversarial Training](https://mlsanity.com/paper/2307.16549/) (2023)
- [Unsupervised word-level prosody tagging for controllable speech synthesis](https://mlsanity.com/paper/2202.07200/) (2022)
- [Acoustic Modeling for End-to-End Empathetic Dialogue Speech Synthesis Using Linguistic and Prosodic Contexts of Dialogue History](https://mlsanity.com/paper/2206.08039/) (2022)
- [Fusion-S2iGan: An Efficient and Effective Single-Stage Framework for Speech-to-Image Generation](https://mlsanity.com/paper/2305.10126/) (2023)
- [BitTTS: Highly Compact Text-to-Speech Using 1.58-bit Quantization and Weight Indexing](https://mlsanity.com/paper/2506.03515/) (2025)
- [Expressive TTS Driven by Natural Language Prompts Using Few Human Annotations](https://mlsanity.com/paper/2311.01260/) (2023)

## Abstract (arXiv)

Generating natural speech with a diverse and smooth prosody pattern is a challenging task. Although random sampling with phone-level prosody distribution has been investigated to generate different prosody patterns, the diversity of the generated speech is still very limited and far from what can be achieved by humans. This is largely due to the use of uni-modal distribution, such as single Gaussian, in the prior works of phone-level prosody modelling. In this work, we propose a novel approach that models phone-level prosodies with a GMM-based mixture density network(MDN) and then extend it for multi-speaker TTS using speaker adaptation transforms of Gaussian means and variances. Furthermore, we show that we can clone the prosodies from a reference speech by sampling prosodies from the Gaussian components that produce the reference prosodies. Our experiments on LJSpeech and LibriTTS dataset show that the proposed method with GMM-based MDN not only achieves significantly better diversity than using a single Gaussian in both single-speaker and multi-speaker TTS, but also provides better naturalness. The prosody cloning experiments demonstrate that the prosody similarity of the proposed method with GMM-based MDN is comparable to recent proposed fine-grained VAE while the target speaker similarity is better.