Multiresolution sinusoidal modeling for wideband audio with modifications

Abstract

In this paper, we describe an computationally efficicnt method of generating more accurate sinusoidal parameters { untplirude, frequency, phase} from a wideband polyphonic audio sourcc in a multiresolution, non-ailiased fashion. This significantly improves upon previous work of sinusoidal modeling that assumes a singlcpitched monophonic source, such as speech or an individual musical instrument, while using approximately the same number of sinusoids. In addition to a more general analysis, we can now pcrform high-quality modifications such as time-stretching and pitchshifting on polyphonic audio with ease.

A Multichannel Sinusoidal Model Applied to Spot Microphone Signals for Immersive Audio

“…For addressing problem 3), use of damped sinusoids and AM modulated sinusoids (instead of constant amplitude sinusoids) has been proposed (e.g., [32], [33]). A multiresolution analysis method [34] has also been proposed for better estimating the sinusoidal parameters by passing the signal through an octave-spaced filterbank, combined with transform coding of transient components. The noise is modeled based on a Bark-band noise model, as in [35].…”

Multichannel Audio Coding for Multimedia Services in Intelligent Environments

“…In music, a harmonic plus noise model was first proposed in [34], where the noise part was modeled based on a piecewise-linear approximation of its short-time spectral envelope or alternatively its LPC envelope (assuming white noise excitation during synthesis). More recent is the work in [35], where multiresolution analysis was applied for better estimating the sinusoidal parameters by passing the signal through an octave-spaced filterbank which was designed for avoiding aliasing problems. Regarding the noise part, it was not parametrically modeled for best audio quality.…”

Modeling Spot Microphone Signals using the Sinusoidal Plus Noise Approach

“…In music, a harmonic plus noise model was first proposed in [10]. More recent is the work in [11], where multiresolution analysis was applied for better estimating the sinusoidal parameters. Regarding the noise part, it was not parametrically modeled for best audio quality.…”