Neural audio codecs have been widely adopted in audio-generative tasks because their compact and discrete representations are suitable for both large-language-model-style and regression-based generative models. However, most neural codecs struggle to model out-of-domain audio, resulting in error propagations to downstream generative tasks. In this paper, we first argue that information loss from codec compression degrades out-of-domain robustness. Then, we propose full-band 48~kHz ComplexDec with complex spectral input and output to ease the information loss while adopting the same 24~kbps bitrate as the baseline AuidoDec and ScoreDec. Objective and subjective evaluations demonstrate the out-of-domain robustness of ComplexDec trained using only the 30-hour VCTK corpus.
View on arXiv@article{wu2025_2502.02019,
title={ ComplexDec: A Domain-robust High-fidelity Neural Audio Codec with Complex Spectrum Modeling },
author={ Yi-Chiao Wu and Dejan Marković and Steven Krenn and Israel D. Gebru and Alexander Richard },
journal={arXiv preprint arXiv:2502.02019},
year={ 2025 }
}