Ecole d'ingénieur et centre de recherche en Sciences du numérique

Artificial bandwidth extension using the constant Q transform

Bachhav, Pramod; Todisco, Massimiliano; Mossi, Moctar; Beaugeant, Christophe; Evans, Nicholas

ICASSP 2017, 42nd IEEE International Conference on Acoustics, Speech and Signal Processing, March 5-9, 2017, New Orleans, USA

Most artificial bandwidth extension (ABE) algorithms are based on the classical source-filter model of speech production. This approach generally requires the dual extension of each component through independent processing. Alternative approaches reported recently operate on the spectrum. With human perception thought to be largely insensitive to phase, most such approaches focus on the extension of the magnitude spectrum alone and rely on Fourier spectral analysis. This paper reports an approach to ABE based on the constant Q transform (CQT), a more perceptually motivated approach to spectral analysis. A Gaussian mixture model is used to estimate missing highband components from available narrowband components before resynthesis with phase estimates obtained from the upsampled narrowband signal. Objective assessment shows that energy normalisation is critical to performance. These findings and the appeal of CQT for ABE are confirmed through informal subjective tests based on the mean opinion score.

Document Doi Bibtex

Titre:Artificial bandwidth extension using the constant Q transform
Mots Clés:bandwidth extension, constant Q transform
Type:Conférence
Langue:English
Ville:New Orleans
Pays:ÉTATS-UNIS
Date:
Département:Sécurité numérique
Eurecom ref:5107
Copyright: © 2017 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.
Bibtex: @inproceedings{EURECOM+5107, doi = {http://dx.doi.org/10.1109/ICASSP.2017.7953218}, year = {2017}, title = {{A}rtificial bandwidth extension using the constant {Q} transform}, author = {{B}achhav, {P}ramod and {T}odisco, {M}assimiliano and {M}ossi, {M}octar and {B}eaugeant, {C}hristophe and {E}vans, {N}icholas}, booktitle = {{ICASSP} 2017, 42nd {IEEE} {I}nternational {C}onference on {A}coustics, {S}peech and {S}ignal {P}rocessing, {M}arch 5-9, 2017, {N}ew {O}rleans, {USA} }, address = {{N}ew {O}rleans, {\'{E}}{TATS}-{UNIS}}, month = {03}, url = {http://www.eurecom.fr/publication/5107} }
Voir aussi: