Exploring speech style spaces with language models: Emotional TTS without emotion labels

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Chandra, Shreeram Suresh, Du, Zongyang, Sisman, Berrak
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909207011262464
author Chandra, Shreeram Suresh
Du, Zongyang
Sisman, Berrak
author_facet Chandra, Shreeram Suresh
Du, Zongyang
Sisman, Berrak
contents Many frameworks for emotional text-to-speech (E-TTS) rely on human-annotated emotion labels that are often inaccurate and difficult to obtain. Learning emotional prosody implicitly presents a tough challenge due to the subjective nature of emotions. In this study, we propose a novel approach that leverages text awareness to acquire emotional styles without the need for explicit emotion labels or text prompts. We present TEMOTTS, a two-stage framework for E-TTS that is trained without emotion labels and is capable of inference without auxiliary inputs. Our proposed method performs knowledge transfer between the linguistic space learned by BERT and the emotional style space constructed by global style tokens. Our experimental results demonstrate the effectiveness of our proposed framework, showcasing improvements in emotional accuracy and naturalness. This is one of the first studies to leverage the emotional correlation between spoken content and expressive delivery for emotional TTS.
format Preprint
id arxiv_https___arxiv_org_abs_2405_11413
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Exploring speech style spaces with language models: Emotional TTS without emotion labels
Chandra, Shreeram Suresh
Du, Zongyang
Sisman, Berrak
Audio and Speech Processing
Machine Learning
Many frameworks for emotional text-to-speech (E-TTS) rely on human-annotated emotion labels that are often inaccurate and difficult to obtain. Learning emotional prosody implicitly presents a tough challenge due to the subjective nature of emotions. In this study, we propose a novel approach that leverages text awareness to acquire emotional styles without the need for explicit emotion labels or text prompts. We present TEMOTTS, a two-stage framework for E-TTS that is trained without emotion labels and is capable of inference without auxiliary inputs. Our proposed method performs knowledge transfer between the linguistic space learned by BERT and the emotional style space constructed by global style tokens. Our experimental results demonstrate the effectiveness of our proposed framework, showcasing improvements in emotional accuracy and naturalness. This is one of the first studies to leverage the emotional correlation between spoken content and expressive delivery for emotional TTS.
title Exploring speech style spaces with language models: Emotional TTS without emotion labels
topic Audio and Speech Processing
Machine Learning
url https://arxiv.org/abs/2405.11413