Editing Spectral band replication (section)

== Methods ==
Encoding of SBR produces a downsampled (usually 2:1) audio signal and guidance information. In an early publication, the guiding data is described as being produced by [[quadrature mirror filter]] (QMF) analysis and an [[envelope estimator]].<ref>{{cite web |last1=Ekstrand |first1=Per |title=Bandwidth extension of audio signals by spectral band replication |url=https://www.esat.kuleuven.be/psi/spraak/seminars/mpca/papers/ekstrand:mpca02.pdf |website=Proc.1st IEEE Benelux Workshop on Model based Processing and Coding of Audio (MPCA-2002), Leuven, Belgium |date=November 2022}}</ref>

Decoding of SBR requires transposing harmonics, a case of [[audio time stretching and pitch scaling]].<ref>{{cite web |last1=Zhong |first1=Haishan |last2=Villemoes |first2=Lars |last3=Ekstrand |first3=Per |last4=Disch |first4=Sascha |last5=Nagel |first5=Frederik |last6=Wilde |first6=Stephan |last7=Chong |first7=Kok Seng |last8=Norimatsu |first8=Takeshi |title=QMF Based Harmonic Spectral Band Replication |url=https://www.aes.org/e-lib/browse.cfm?elib=16043 |publisher=Audio Engineering Society |language=English |date=19 October 2011}}</ref>
* A traditional approach starts with small intervals of [[discrete fourier transform]] (DFT), phase adjustments, IDFT, and ends with overlap-add. This method is sensitive to transient signals which can cause echos, requiring some padding (50% in USAC) in the DFT.
* A newer approach is the QMF. One single filter bank can perform a whole time-stretch and pitch-scale operation for lower computational complexity.