In this page, we demonstrate our proposed discrete diffusion-based piano accompaniment generation model, D3PIA, leveraging the locally aligned structure of musical accompaniments with the lead sheet in the piano roll representation. As mentioned in the Results & Discussions Section, we also present the objective and subjective scores of samples.
Trained with 8 bars. Temperature = 1.5
Test split of POP909 Dataset including melody, bridge, and arrangement track
Trained with 8 bars. Generated with inpainting "below" option. chord CFG cond_scale = 5.0
Re-implemented (GPT-2) embellish part. Temp = 1.1, p = 0.95
Stage 4 model of WholeSongGen.
Trained with 8 bars (4 bars in original code).
Melody and chord. Segments selected when melody exists in ≥4 bars.
| Objective Metrics | Subjective Metrics | |||
|---|---|---|---|---|
| Model | Chord Accuracy | Chord Similarity | Harmony | Correctness |