Neural Sample-based Piano Synthesis

University of Oslo

Abstract

Piano sound synthesis has been an active topic of research and development for several decades. Although comprehensive physics-based piano models have been proposed, sample-based piano emulation is still widely utilized for its computational efficiency and relative accuracy despite presenting significant memory storage requirements. This paper proposes a novel hybrid approach to sample-based piano synthesis aimed at improving the fidelity of sound emulation while reducing memory requirements for storing samples. A neural network-based model processes the sound recorded from a single piano key at a given velocity. The network is trained to learn the nonlinear relationship between the various velocities at which a piano key is pressed and the corresponding sound alterations. Results show that the method achieves high accuracy using a specific neural architecture that is computationally efficient, presenting few trainable parameters, and it requires memory only for one sample for each piano key.

BiViB Dataset - Upright Piano

Piano-specific model

Parameters Real Model

Key: A0

Key: B1

Key: C2

Key: D3

Key: E4

Key: F5

Key: G6

Key: A#7

BiViB Dataset - Upright Piano

Scenario T2: the model is trained with the recording with maximum velocity as input. The examples refer to a velocity of 23.

Parameters Real Model

Key: A0

Key: B1

Key: C2

Key: D3

Key: E4

Key: F5

Key: G6

Key: A#7

BiViB Dataset - Upright Piano

Scenario T1: the model is trained with the recording with the lowest velocity as input. The examples refer to a velocity of 100.

Parameters Real Model

Key: A0

Key: B1

Key: C2

Key: D3

Key: E4

Key: F5

Key: G6

Key: A#7

BibTeX


@inproceedings{simionato_neuralpcm_2025,
	author = {Simionato Riccardo and Fasciani Stefano},
	title = {Neural Sampled-based Piano Synthesis},
	booktitle = {Proceedings of the International Conference on Digital Audio Effects (DAFx25)},
	year = {2025},
    address = {Ancona, Italy},
}