Morphing Musical Instrument Sounds with the Sinusoidal Model in the Sound Morphing Toolbox

Abstract

Sound morphing stands out among the sound transformation techniques in the literature due to its creative and research potential. The aim of sound morphing is to gradually blur the categorical distinction between the source and target sounds by blending sensory attributes. As such, the focus and ultimate challenge of most sound morphing techniques is to interpolate across dimensions of timbre perception to achieve the desired result. There are several sound morphing proposals in the literature with few open-source implementations freely available, making it difficult to reproduce the results, compare models, or simply use them in other applications such as music composition, sound design, and timbre research. This work describes how to morph musical instrument sounds with the sinusoidal model using the sound morphing toolbox (SMT), a freely available and open-source piece of software. The text describes the audio processing steps required to morph sounds with the SMT using a step-by-step example to illustrate the need for and the result of each step. The SMT contains implementations of a sound morphing algorithm in MATLAB ® that were designed to be as easy as possible to understand and use, giving the user control over the result and full customization.

Interpolation of Synthesizer Presets using Timbre-Regularized Auto-Encoders

“…Sound morphing consists in the hybridization of two input audio waveforms. The most common method is based on sinusoidal modeling of the input signals, followed by interpolation of the time-varying parameters (frequencies and amplitudes of the partials) and additive resynthesis [32]- [36]. The non-harmonic parts of input signals, e.g., transients or noisy sounds which cannot be easily reproduced by pure additive synthesis, can be handled separately.…”

“…It prevents their adoption by users who choose specific synthesizers for their characteristic audio qualities. Moreover, dedicated sound models often yield suboptimal results with different acoustic sources [36].…”

“…For instance, [39] performs resynthesis from smooth time-frequency representations, and [10], [40] compute interpolations of smoothed spectral envelopes. Morphing through additive resynthesis [36] using matched partials is also inherently smooth. In contrast, widespread synthesizers typically provide some non-differentiable and non-linear In this work, the goal is to build a model that enables interpolation of presets for any synthesizer (Figure 2).…”

“…Hence, an automatic, explainable and reproducible method is desirable. A general framework [70] suggests to use criteria such as correspondence, intermediateness and smoothness. The same authors [10] eventually evaluate the quality of a morphing by computing the non-linearity of four spectral and two temporal features.…”

A perceptual sound space for auditory displays based on sung-vowel synthesis

“…In this way, two opposite timbres on the circle have maximal perceptual distance. However, seamless transitions between neighboring timbres are possible only by some form of audio timbre morphing 40 . This would be important to achieve a space where timbre is globally categorical yet locally continuous, similarly to the hue in the HSL color space.…”