Multiple Description Spherical Quantization of Sinusoidal Parameters with Repetition Coding of the Amplitudes

Publikation: Bidrag til tidsskriftKonferenceartikel i tidsskriftForskningpeer review

103 Downloads (Pure)

Resumé

Recently, multiple description spherical trellis-coded quantization (MDSTCQ) for quantization of sinusoidal parameters
was proposed, which suffered from a suboptimal implementation. Therefore, we propose the multiple description spherical quantization with repetition coding of the amplitudes (MDSQRA) scheme. Here, we apply repetition coding to the amplitudes, whereas the phases and the frequencies are coded using multiple description quantization. We measure the relation between the expected perceptual distortions for the two quantization schemes, which shows that MDSQRA outperforms MDSTCQ for most packet-loss probabilities. Also, we measure the perceptual distortion obtained when applying MDSQRA on synthetic data and show that it matches the expected perceptual distortion. Furthermore, due to MDSTCQs suboptimal implementation, MDSQRA outperforms MDSTCQ in terms of measured perceptual distortion even at packet-loss probabilities where the expected perceptual distortion was lowest for MDSTCQ.


OriginalsprogEngelsk
TidsskriftProc. of Asilomar Conference on Signals, Systems, and Computers
Antal sider5
ISSN1058-6393
StatusUdgivet - 2009
BegivenhedRec. Asilomar Conf. Signals, Systems, and Computers - Monterey, CA, USA
Varighed: 1 nov. 20094 nov. 2009

Konference

KonferenceRec. Asilomar Conf. Signals, Systems, and Computers
LandUSA
ByMonterey, CA
Periode01/11/200904/11/2009

Fingerprint

Packet loss

Citer dette

@inproceedings{360609009aef11deb96d000ea68e967b,
title = "Multiple Description Spherical Quantization of Sinusoidal Parameters with Repetition Coding of the Amplitudes",
abstract = "Recently, multiple description spherical trellis-coded quantization (MDSTCQ) for quantization of sinusoidal parameterswas proposed, which suffered from a suboptimal implementation. Therefore, we propose the multiple description spherical quantization with repetition coding of the amplitudes (MDSQRA) scheme. Here, we apply repetition coding to the amplitudes, whereas the phases and the frequencies are coded using multiple description quantization. We measure the relation between the expected perceptual distortions for the two quantization schemes, which shows that MDSQRA outperforms MDSTCQ for most packet-loss probabilities. Also, we measure the perceptual distortion obtained when applying MDSQRA on synthetic data and show that it matches the expected perceptual distortion. Furthermore, due to MDSTCQs suboptimal implementation, MDSQRA outperforms MDSTCQ in terms of measured perceptual distortion even at packet-loss probabilities where the expected perceptual distortion was lowest for MDSTCQ.",
author = "Jensen, {Jesper Rindom} and Christensen, {Mads Gr{\ae}sb{\o}ll} and Larsen, {Morten Holm} and Jensen, {S{\o}ren Holdt} and Torben Larsen",
year = "2009",
language = "English",
journal = "Asilomar Conference on Signals, Systems and Computers. Conference Record",
issn = "1058-6393",
publisher = "I E E E Computer Society",

}

TY - GEN

T1 - Multiple Description Spherical Quantization of Sinusoidal Parameters with Repetition Coding of the Amplitudes

AU - Jensen, Jesper Rindom

AU - Christensen, Mads Græsbøll

AU - Larsen, Morten Holm

AU - Jensen, Søren Holdt

AU - Larsen, Torben

PY - 2009

Y1 - 2009

N2 - Recently, multiple description spherical trellis-coded quantization (MDSTCQ) for quantization of sinusoidal parameterswas proposed, which suffered from a suboptimal implementation. Therefore, we propose the multiple description spherical quantization with repetition coding of the amplitudes (MDSQRA) scheme. Here, we apply repetition coding to the amplitudes, whereas the phases and the frequencies are coded using multiple description quantization. We measure the relation between the expected perceptual distortions for the two quantization schemes, which shows that MDSQRA outperforms MDSTCQ for most packet-loss probabilities. Also, we measure the perceptual distortion obtained when applying MDSQRA on synthetic data and show that it matches the expected perceptual distortion. Furthermore, due to MDSTCQs suboptimal implementation, MDSQRA outperforms MDSTCQ in terms of measured perceptual distortion even at packet-loss probabilities where the expected perceptual distortion was lowest for MDSTCQ.

AB - Recently, multiple description spherical trellis-coded quantization (MDSTCQ) for quantization of sinusoidal parameterswas proposed, which suffered from a suboptimal implementation. Therefore, we propose the multiple description spherical quantization with repetition coding of the amplitudes (MDSQRA) scheme. Here, we apply repetition coding to the amplitudes, whereas the phases and the frequencies are coded using multiple description quantization. We measure the relation between the expected perceptual distortions for the two quantization schemes, which shows that MDSQRA outperforms MDSTCQ for most packet-loss probabilities. Also, we measure the perceptual distortion obtained when applying MDSQRA on synthetic data and show that it matches the expected perceptual distortion. Furthermore, due to MDSTCQs suboptimal implementation, MDSQRA outperforms MDSTCQ in terms of measured perceptual distortion even at packet-loss probabilities where the expected perceptual distortion was lowest for MDSTCQ.

M3 - Conference article in Journal

JO - Asilomar Conference on Signals, Systems and Computers. Conference Record

JF - Asilomar Conference on Signals, Systems and Computers. Conference Record

SN - 1058-6393

ER -