TQCodec: Towards neural audio codec for high-fidelity music streaming

2026-03-02

AI Summary

This work addresses the limitations of existing neural audio codecs, which primarily target ultra-low bitrates (≤16 kbps) and fail to meet the high-fidelity requirements of music streaming at 32–128 kbps. To bridge this gap, we propose TQCodec, a lightweight, asymmetric codec based on the SEANet architecture that supports 44.1 kHz sampling. TQCodec incorporates SimVQ vector quantization to preserve mid-frequency details, a phase-aware waveform loss function to enhance reconstruction accuracy, and an auditory-perception-driven bit allocation strategy that prioritizes perceptually critical low-frequency components. Experimental results across multiple music datasets demonstrate that TQCodec significantly outperforms current methods within the target bitrate range, achieving high-quality audio reconstruction suitable for high-fidelity streaming applications.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

We propose TQCodec, a neural audio codec designed for high-bitrate, high-fidelity music streaming. Unlike existing neural codecs that primarily target ultra-low bitrates (<= 16kbps), TQCodec operates at 44.1 kHz and supports bitrates from 32 kbps to 128 kbps, aligning with the standard quality of modern music streaming platforms. The model adopts an encoder-decoder architecture based on SEANet for efficient on-device computation and introduces several enhancements: an imbalanced network design for improved quality with low overhead, SimVQ for mid-frequency detail preservation, and a phase-aware waveform loss. Additionally, we introduce a perception-driven band-wise bit allocation strategy to prioritize perceptually critical lower frequencies. Evaluations on diverse music datasets demonstrate that TQCodec achieves superior audio quality at target bitrates, making it well-suited for high-quality audio applications.

Problem

• neural audio codec
• high-fidelity music streaming
• bitrate
• audio quality
• music compression

Innovation

• neural audio codec
• high-fidelity music streaming
• perception-driven bit allocation
• SimVQ
• phase-aware waveform loss