Discrete-time signal processing (2nd ed.)
Discrete-time signal processing (2nd ed.)
Multiple fundamental frequency estimation and polyphony inference of polyphonic music signals
IEEE Transactions on Audio, Speech, and Language Processing
Tracking of partials for additive sound synthesis using hidden Markov models
ICASSP'93 Proceedings of the 1993 IEEE international conference on Acoustics, speech, and signal processing: plenary, special, audio, underwater acoustics, VLSI, neural networks - Volume I
New Polyphonic Sound Generator Chip with Integrated Microprocessor-Programmable ADSR Envelope Shaper
IEEE Transactions on Consumer Electronics
| Hi-index | 0.00 |
In this paper, a method for automatic reduction of dissonance in recorded isolated chords is proposed. Previous approaches address this problem using source separation and note-level processing. In our approach, we manipulate the harmonic structure as a whole in order to avoid beating partials which, according to prior research on dissonance perception, typically produce an unpleasant sound. The proposed system firstly performs a sinusoidal plus residual modeling of the input and analyses the various fundamental frequencies present in the chord. This information is used to create a symbolic representation of the in-tune version of the input according to some musical rules. Then, the partials of the signals are shifted in order to fit the in-tune harmonic structure of the input chord. The input is assumed to contain one isolated chord, with relatively stable fundamental frequencies belonging to the Western chromatic scale. The evaluation has been performed by 31 expert musicians, which have quantified the perceived consonance of six varied, out-of-tune chords in three variants: unprocessed, processed with our system and processed by a state-of-the-art commercial tool (Melodyne Editor). The proposed approach attains an important reduction of the perceived dissonance, showing better performance than Melodyne Editor for most of the cases evaluated.