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

TL;DR

This paper introduces a novel emotional TTS system that leverages GPT-3 to predict emotion attributes directly from text, enabling paragraph-level emotional speech synthesis without auxiliary inputs.

Contribution

It presents a GPT-3 based emotion prediction method that estimates emotion class and strength directly from text, improving emotional speech synthesis.

Findings

Effective paragraph-level emotional speech generation

No need for auxiliary emotion labels or classes

Enhanced emotional context understanding

Abstract

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.