SSR-Speech: Towards Stable, Safe and Robust Zero-shot Text-based Speech Editing and Synthesis

2024-09-11

arXiv.org

AI Summary

This work addresses key challenges in zero-shot text-driven speech editing and TTS synthesis—namely, poor stability, weak security, limited robustness to interference, and difficulty in multi-segment editing. To this end, we propose a novel neural auto-regressive encoder-decoder model. Methodologically, it adopts a Transformer-based decoder architecture with classifier-free guidance; introduces a frame-level detectable watermark encoder for precise localization of edited regions; reconstructs waveforms by fusing original speech segments to enhance fidelity; and incorporates a customized Watermark Encodec for improved robustness. Experiments on RealEdit and LibriTTS benchmarks demonstrate state-of-the-art performance: the model enables fine-grained, multi-segment editing, exhibits strong robustness against background noise, and ensures both content security and edit traceability.

Technology Category

• Natural Language Processing: Safety and Robustness
• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Diffusion Models for Vision

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

In this paper, we introduce SSR-Speech, a neural codec autoregressive model designed for stable, safe, and robust zero-shot textbased speech editing and text-to-speech synthesis. SSR-Speech is built on a Transformer decoder and incorporates classifier-free guidance to enhance the stability of the generation process. A watermark Encodec is proposed to embed frame-level watermarks into the edited regions of the speech so that which parts were edited can be detected. In addition, the waveform reconstruction leverages the original unedited speech segments, providing superior recovery compared to the Encodec model. Our approach achieves state-of-the-art performance in the RealEdit speech editing task and the LibriTTS text-to-speech task, surpassing previous methods. Furthermore, SSR-Speech excels in multi-span speech editing and also demonstrates remarkable robustness to background sounds. The source code and demos are released.

Problem

• Speech Editing
• Text-to-Speech Conversion
• Robustness to Interference

Innovation

• SSR-Speech
• Transformer Decoder
• Watermarking Technique