Newest papers

Piano2Pop: Expanding Piano Demos into Pop Productions Without Aligned Supervision

Yun-Chen Cheng, Chih-Pin Tan, Tianda Huang, Yi-Hsuan Yang · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.31510 · cited by 0 works (OpenAlex) · open-access PDF

Korean

피아노2팝(Piano2Pop)는 피아노 미디(MIDI)를 팝 녹음에서 유래한 합성 피아노 도구 쌍으로 변환하는 음악 확장 작업이다. 음악 생성과 폴리폰 음악 전송의 최근 진보에 의해 가능하게 되면서, 피아노 롤에 대한 프리트레이닝 플로우 매칭 모델을 프리트레이닝할 수 있다. 우리는 팝 녹음에서 유래한 합성 피아노 도구 쌍을 훈련하고, 기능 선택 (FST) 및 레지스트리 밀도 (RDA) 증폭을 포함한 쌍 무료 전송을 비교하며, 쌍을 감독하는 피아노 덮개 생성기를 pseudo-demo 소스로 제공합니다. 476에서 실제 피아노 덮개 / 팝 쌍을

Machine translation (Meta M2M100); the English original follows.

Abstract

We introduce Piano2Pop, a music-expansion task that transforms piano MIDI into instrumental pop productions. Enabled by recent advances in music generation and polyphonic music transcription, our pseudo-supervised approach conditions a pretrained flow-matching model on piano rolls without aligned piano-pop supervision. We train on synthetic piano-instrumental pairs derived from pop recordings, comparing pair-free transcription, including function-selective (FST) and register-density (RDA) augmentations, with a pair-supervised piano-cover generator as pseudo-demo sources. On 476 testing real-world piano-cover/pop pairs, the pair-supervised approach achieves the best Frechet Audio Distance, Accompaniment Prompt Adherence, and spectral shape on real-piano inputs, while FST mitigates piano-heavy outputs in pair-free training and remains a promising alternative. All fine-tuned systems significantly outperform the zero-shot baselines, which suggests a role for task-specific modeling.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Piano, MIDI, Polyphony, Speech recognition, Generator (circuit theory), Computer science

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange