Benchmarking Language Modeling for Lossless Compression of Full-Fidelity Audio
For audio compression researchers, this work extends LM-based lossless compression to practical bit depths but reveals diminishing returns, highlighting the need for further improvements.
The paper benchmarks language model-based lossless audio compression on full-fidelity audio (16/24-bit), proposing Trilobyte tokenization to handle high bit depths. LMs outperform FLAC at 8 and 16-bit but show diminishing gains at higher bit depths.
Autoregressive "language" models (LMs) trained on raw waveforms can be repurposed for lossless audio compression, but prior work is limited to 8-bit audio, leaving open whether such approaches work for practical settings (16/24-bit) and can compete with existing codecs. We benchmark LM-based compression on full-fidelity audio across diverse domains (music, speech, bioacoustics), sampling rates (16kHz-48kHz), and bit depths (8, 16, 24-bit). Standard sample-level tokenization becomes intractable at higher bit depths due to vocabulary size (65K for 16-bit; 16.7M for 24-bit). We propose Trilobyte, a byte-level tokenization schema for full resolution audio, improving vocabulary scaling from $O(2^{b})$ to $O(1)$ and enabling the first tractable 24-bit LM-based lossless compression. While LMs consistently outperform FLAC and yield state-of-the-art compression at 8-bit and 16-bit, we observe that compression gains become more modest as bit depth increases beyond 8-bit.