Learnable Classifier-Free Guidance Null Embeddings for Enhanced Controllable Speech Synthesis

2026-09-21

AI Summary

本文通过引入可学习的无条件嵌入替代固定空向量，增强可控语音合成的质量和鲁棒性，同时提供对说话人和文本引导的精细控制。

Abstract

Classifier-free Guidance (CFG) is widely adopted in text-to-speech (TTS) systems to enhance generation quality and conditioning fidelity by interpolating between conditioned and unconditioned predictions. A common unconditional technique is to use an empty representation, in the form of a fixed null vector. In this work, we propose replacing this representation with a learnable unconditional embedding, optimized to represent a meaningful unconditional state. Objective and subjective evaluations demonstrate that learnable null embeddings consistently outperform fixed null embeddings across speaker similarity, speech stability, and expressiveness, while exhibiting greater robustness to larger guidance scales. We further show that learning a distinct unconditional embedding for each of the TTS conditioning modalities allows fine-grained control over speaker and text guidance, showcasing the trade-off between similarity and quality, and stability and expressiveness in the generated speech.

Problem

• Classifier-free Guidance
• Text-to-Speech
• Controllable Speech Synthesis
• Unconditional Embedding

Innovation

• learnable null embeddings
• classifier-free guidance
• text-to-speech
• unconditional embedding
• fine-grained control