On perceptual distortion minimization and nonlinear least-squares frequency estimation

Christensen, Mads Græsbøll; Jensen, Søren Holdt

doi:10.1109/tsa.2005.860347

Cited by 25 publications

(36 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For an overview of methods for modeling an audio segments using sinusoids such that (1) is minimized, we refer the interested reader to [14] and the references therein.…”

Section: Spherical Trellis-coded Quantizationmentioning

confidence: 99%

Variable Dimension Trellis-Coded Quantization of Sinusoidal Parameters

Larsen

Christensen

Jensen

2008

IEEE Signal Process. Lett.

View full text Add to dashboard Cite

Abstract-In this letter, we propose joint quantization of the parameters of a set of sinusoids based on the theory of trellis-coded quantization. A particular advantage of this approach is that it allows for joint quantization of a variable number of sinusoids, which is particularly relevant in variable rate parametric audio coding. Under high-resolution assumptions and based on a perceptually relevant distortion measure, we derive analytical expressions for the optimal design subject to an entropy constraint. Numerical experiments show a significant performance gain compared to optimal spherical quantization at the cost of a slight increase in computational complexity.

show abstract

“…For an overview of methods for modeling an audio segments using sinusoids such that (1) is minimized, we refer the interested reader to [14] and the references therein.…”

Section: Spherical Trellis-coded Quantizationmentioning

confidence: 99%

Variable Dimension Trellis-Coded Quantization of Sinusoidal Parameters

Larsen

Christensen

Jensen

2008

IEEE Signal Process. Lett.

View full text Add to dashboard Cite

show abstract

“…It turns out, that for this asymptotic equivalence to hold, W has to have a certain structure. Such a structure is induced by the perceptual distortion measure in [3] which leads to W being a filtering matrix [8,16]. It is quite clear that this is a rather special case.…”

Section: Estimation and Codingmentioning

confidence: 99%

“…Also, computation time is not wasted on extracting sinusoids that are not used anyway. As explained in [16], some of these methods are also asymptotically optimal (for a large number of samples N ) from a statistical point of view.…”

Section: Perception In Estimatorsmentioning

confidence: 99%

“…For example, if we seek to estimate the frequencies of a set of sinusoids in x from y, then the sinusoids have to be invariant to the transformation U. Or, in other words, the vectors of A(θ) have to be eigenvectors, or good approximations thereof, of U [16]. For the distortion measure in [3], this is asymptotically the case for sinusoids, because it induces a certain structure on U.…”

Section: Perception In Estimatorsmentioning

confidence: 99%

“…Some examples of the use of perceptual models in signal processing, aside from the obvious examples of mp3 and AAC, are analysis-by-synthesis speech coding [1], audio coding based on pre-and post-filtering [7], sinusoidal modeling using pre-filtering [8], weighted matching pursuit [9,10], and the psycho-acoustic adaptive matching pursuit [11], pre-filtered auto-regressive noise modeling [12], post-filtering in speech coding [13], perceptual shaping in subspace-based speech enhancement [14], and pre-processing in pitch estimation [15]. Note that the matching pursuit methods mentioned above have been shown to be equivalent under certain conditions in [16]. The application to coding is obvious: a better quality can be obtained at the same bit-rate if the distortion measure used in the encoding process is replaced by one that better resembles the human auditory system, in fact the difference can be quite dramatic, hence the miracle.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

On perceptual distortion measures and parametric modeling

Christensen

2008

The Journal of the Acoustical Society of America

View full text Add to dashboard Cite

Over the past two decades, there has been much interest in incorporating human sound perception in signal processing algorithms. An example of this is MDCT-based audio coding where a good quality can be achieved at very low bit-rates by taking masking effects into account. More recently, the same principles have been applied to parametric modeling and coding of audio signals. We discuss the inherent tradeoffs in choosing a perceptual distortion measure and a parametric model, and the pros and cons of various ways of implementing such perceptual distortion measures are discussed. An important question that we seek to answer is whether perception should be taken into account in the estimation of model parameters or this should be done in a separate step.

show abstract

Parametric Models

2008

Modeling, Estimation and Optimal Filtering in Signal Processing

View full text Add to dashboard Cite

On perceptual distortion minimization and nonlinear least-squares frequency estimation

Cited by 25 publications

References 48 publications

Variable Dimension Trellis-Coded Quantization of Sinusoidal Parameters

Variable Dimension Trellis-Coded Quantization of Sinusoidal Parameters

On perceptual distortion measures and parametric modeling

Parametric Models

Contact Info

Product

Resources

About