arXiv · 2605.15831
Modeling Music as a Time-Frequency Image: A 2D Tokenizer for Music Generation
Abstract
Autoregressive music generation depends strongly on the audio tokenizer. Existing high-fidelity codecs often use residual multi-codebook quantization, which preserves reconstruction quality but complicates language modeling after sequence flattening, as the residual hierarchy imposes strong sequential dependencies and can amplify error accumulation. We propose BandTok, a generation-oriented 2D Mel-spectrogram tokenizer that represents each frame with Mel-frequency band tokens from a single shared codebook. This design yields a physically interpretable time-frequency token grid with a more independent token structure, making it better suited for autoregressive modeling. BandTok improves reconstruction with a multi-scale PatchGAN objective and EMA codebook updates. We further introduce an autoregressive language model with 2D Rotary Position Embedding (2D RoPE) to preserve temporal and frequency-band structure during generation. Experiments show that BandTok improves over residual-codebook tokenizers and achieves strong results in a data-limited setting. The source code and generation demos for this work are publicly available.
Explore related subjects
Keep this discovery
Explore connections, maps & timelines
Yuqing Cheng, Xingyu Ma, Guochen Yu, Xiaotao Gu. 2026-05-15. Modeling Music as a Time-Frequency Image: A 2D Tokenizer for Music Generation. https://arxiv.org/abs/2605.15831
Cite the original work for its findings. Save a collection to share your selection of sources.