# Language Model-Based Emotion Prediction Methods for Emotional Speech Synthesis Systems

arXiv: https://arxiv.org/abs/2206.15067 · HTML: https://mlsanity.com/paper/2206.15067/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 30 Jun 2022.

## Built on by (most cited first, 9 of 9)

- [EmoSphere-TTS: Emotional Style and Intensity Modeling via Spherical Emotion Vector for Controllable Emotional Text-to-Speech](https://mlsanity.com/paper/2406.07803/) (2024)
- [DiffProsody: Diffusion-based Latent Prosody Generation for Expressive Speech Synthesis with Prosody Conditional Adversarial Training](https://mlsanity.com/paper/2307.16549/) (2023)
- [Emotional Dimension Control in Language Model-Based Text-to-Speech: Spanning a Broad Spectrum of Human Emotions](https://mlsanity.com/paper/2409.16681/) (2024)
- [EE-TTS: Emphatic Expressive TTS with Linguistic Information](https://mlsanity.com/paper/2305.12107/) (2023)
- [Exploring speech style spaces with language models: Emotional TTS without emotion labels](https://mlsanity.com/paper/2405.11413/) (2024)
- [Controllable Speaking Styles Using a Large Language Model](https://mlsanity.com/paper/2305.10321/) (2023)
- [NaturalVoices: A Large-Scale, Spontaneous and Emotional Podcast Dataset for Voice Conversion](https://mlsanity.com/paper/2511.00256/) (2025)
- [EmoSpeech: A Corpus of Emotionally Rich and Contextually Detailed Speech Annotations](https://mlsanity.com/paper/2412.06581/) (2024)
- [Improving French Synthetic Speech Quality via SSML Prosody Control](https://mlsanity.com/paper/2508.17494/) (2025)

## Abstract (arXiv)

This paper proposes an effective emotional text-to-speech (TTS) system with a pre-trained language model (LM)-based emotion prediction method. Unlike conventional systems that require auxiliary inputs such as manually defined emotion classes, our system directly estimates emotion-related attributes from the input text. Specifically, we utilize generative pre-trained transformer (GPT)-3 to jointly predict both an emotion class and its strength in representing emotions coarse and fine properties, respectively. Then, these attributes are combined in the emotional embedding space and used as conditional features of the TTS model for generating output speech signals. Consequently, the proposed system can produce emotional speech only from text without any auxiliary inputs. Furthermore, because the GPT-3 enables to capture emotional context among the consecutive sentences, the proposed method can effectively handle the paragraph-level generation of emotional speech.