EmoSpeech: Guiding FastSpeech2 Towards Emotional Text to Speech

TL;DR

EmoSpeech enhances FastSpeech2 for emotional speech synthesis by introducing modifications that improve naturalness and emotional expressiveness, validated through both automatic and human evaluations.

Contribution

The paper presents novel modifications to FastSpeech2, enabling more expressive emotional speech synthesis with improved quality and emotion recognition accuracy.

Findings

EmoSpeech outperforms existing models in MOS and emotion recognition.

The conditioning mechanism effectively manages emotion distribution in speech.

Human evaluations favor EmoSpeech's naturalness and emotional expressiveness.

Abstract

State-of-the-art speech synthesis models try to get as close as possible to the human voice. Hence, modelling emotions is an essential part of Text-To-Speech (TTS) research. In our work, we selected FastSpeech2 as the starting point and proposed a series of modifications for synthesizing emotional speech. According to automatic and human evaluation, our model, EmoSpeech, surpasses existing models regarding both MOS score and emotion recognition accuracy in generated speech. We provided a detailed ablation study for every extension to FastSpeech2 architecture that forms EmoSpeech. The uneven distribution of emotions in the text is crucial for better, synthesized speech and intonation perception. Our model includes a conditioning mechanism that effectively handles this issue by allowing emotions to contribute to each phone with varying intensity levels. The human assessment indicates that proposed modifications generate audio with higher MOS and emotional expressiveness.