VoCodec: An Efficient Lightweight Low-Bitrate Speech Codec

TL;DR

VoCodec tackles real-time speech coding at very low bitrates under tight computational budgets. It combines a time-frequency domain encoder/decoder based on a Vocos backbone with Residual Vector Quantization, and optionally a front-end UL-UNAS for noise and reverberation suppression. The method achieves low latency and competitive objective and subjective performance, ranking fourth in LRAC Track 1 and attaining top MUSHRA on clean speech, with robustness to noise and reverberation in the cascaded setup. This work demonstrates practical, high-quality speech reconstruction under resource constraints and points to future improvements at ultra-low bitrates.

Abstract

Recent advancements in end-to-end neural speech codecs enable compressing audio at extremely low bitrates while maintaining high-fidelity reconstruction. Meanwhile, low computational complexity and low latency are crucial for real-time communication. In this paper, we propose VoCodec, a speech codec model featuring a computational complexity of only 349.29M multiply-accumulate operations per second (MACs/s) and a latency of 30 ms. With the competitive vocoder Vocos as its backbone, the proposed model ranked fourth on Track 1 in the 2025 LRAC Challenge and achieved the highest subjective evaluation score (MUSHRA) on the clean speech test set. Additionally, we cascade a lightweight neural network at the front end to extend its capability of speech enhancement. Experimental results demonstrate that the two systems achieve competitive performance across multiple evaluation metrics. Speech samples can be found at https://acceleration123.github.io/.

Figures (1)

• Figure 1: The architecture of the proposed model