DuET jointly transcribes piano notes and tracks beats for scores
Original title: DuET: Note-Frame Dual Encoding for Piano Performance-to-Score Transcription
On Gangqinpu, DuET cuts MUSTER mean error from 6.56% to 3.64% versus the strongest baseline.
DuET is a Dual-Encoder Transformer that jointly learns note-level performance MIDI-to-score transcription and frame-level beat and downbeat tracking. Bidirectional cross-attention links note and frame encoders, and an autoregressive decoder predicts structured score attributes. TimeRoPE and TPRoPE encode a shared performance-time reference plus relative time and pitch. Code, weights, and demos are released.
- Evaluations use the ASAP classical piano dataset and the curated Gangqinpu dataset.
- On ASAP, MUSTER mean error falls from 11.30% to 11.22%.
- Ablations support frame-level context, time-based positional encoding, and decoder-history masking.
- Code and weights are at github.com/wei-zeng98/DuET.
From the original article
1 min read at huggingface.co
Transcribing expressive piano performances into scores requires interpreting individual note timings within an underlying metrical structure. However, score attributes are associated with notes, whereas beats may occur without coincident note onsets. We introduce DuET, a Dual-Encoder Transformer that jointly learns note-level performance MIDI-to-score (PM2S) transcription and frame-level beat and downbeat tracking.
By Wei Zeng, Ye Wang · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.