DS-Codec：基于镜像到非镜像架构切换的双阶段训练语音编解码器
DS-Codec: Dual-Stage Training with Mirror-to-NonMirror Architecture Switching for Speech Codec
AI总结：
本文提出DS-Codec，一种采用镜像与非镜像架构切换的双阶段训练神经语音编解码器，通过平衡两者优势以增强码本鲁棒性并实现高保真语音重建。
摘要：
神经语音编解码器对于推进文本到语音（TTS）系统至关重要。随着大语言模型在文本生成领域的近期成功，开发高质量的语音分词器变得愈发重要。本文提出了DS-Codec，一种新型神经语音编解码器，其采用具有镜像与非镜像架构切换的双阶段训练框架，旨在实现卓越的语音重建。我们进行了广泛的实验和消融研究，以评估该训练策略的有效性，并比较了两种架构的性能。结果表明，镜像结构显著增强了所学码本的鲁棒性，且该训练策略平衡了镜像与非镜像结构间的优势，从而改进了高保真语音重建。
英文摘要：
Neural speech codecs are essential for advancing text-to-speech (TTS) systems. With the recent success of large language models in text generation, developing high-quality speech tokenizers has become increasingly important. This paper introduces DS-Codec, a novel neural speech codec featuring a dual-stage training framework with mirror and non-mirror architectures switching, designed to achieve superior speech reconstruction. We conduct extensive experiments and ablation studies to evaluate the effectiveness of our training strategy and compare the performance of the two architectures. Our results show that the mirrored structure significantly enhances the robustness of the learned codebooks, and the training strategy balances the advantages between mirrored and non-mirrored structures, leading to improved high-fidelity speech reconstruction.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
