Spectral Band Replication: Difference between revisions

From MultimediaWiki
Jump to navigation Jump to search
m (fix subheading link)
No edit summary
 
(5 intermediate revisions by one other user not shown)
Line 1: Line 1:
Spectral Band Replication or SBR is a lossy scheme developed by Coding Technologies (now part of [[Dolby Technologies|Dolby]]) to efficiently reconstruct the high frequency data of an audio signal from the low frequency data. SBR is paired with a lossy audio codec (known as the core coder) at running half the target sample rate. Each audio frame is decoded like normal. The decoded samples are then pulled into the SBR decoder along with encoded SBR side information and the high frequency data is generated and recombined with the low frequency data. The full spectral data is then outputted at the target sample rate. Thanks to the standardization of of HE-AAC, technical details are not available from ISO.
*Specification: 14496-3:3009
**[http://www.3gpp.org/ftp/Specs/archive/26_series/26.404/26404-900.zip 26.404-9.0.0] (free)
 
Spectral Band Replication or SBR is a lossy scheme developed by Coding Technologies (now part of [[Dolby Technologies|Dolby]]) to efficiently reconstruct the high frequency data of an audio signal from the low frequency data. SBR is paired with a lossy audio codec (known as the core coder) at running half the target sample rate. Each audio frame is decoded like normal. The decoded samples are then pulled into the SBR decoder along with encoded SBR side information and the high frequency data is generated and recombined with the low frequency data. The full spectral data is then outputted at the target sample rate. Thanks to the standardization of of HE-AAC, technical details are available from ISO.
 
===SBR QMF filterbank===
At the heart of SBR lies a 32 analysis filterbank and an 64 band synthesis filterbank. The SBR QMF is also used by [[Parametric Stereo]] and [[MPEG Surround]].
 
A description of the SBR filterbank can be found in 26.404.


===Codecs Using SBR===
===Codecs Using SBR===
*[[MP3#mp3PRO|MP3pro]] - SBR with [[MP3]]
*[[MP3#mp3PRO|MP3pro]] - SBR with [[MP3]]
*[[AAC#The HE-AAC Profile (v1 and v2)|HE-AAC]] - SBR with [[AAC]] (also known as aacPlus)
*[[AAC#The HE-AAC Profile (v1 and v2)|HE-AAC]] - SBR with [[AAC]] (also known as aacPlus)
*[[Dolby Pulse]] - HE-AAC with Dolby metadata


===Links===
===Links===
*http://www.codingtechnologies.com/products/sbr.htm
*http://web.archive.org/web/20071021030240/http://www.codingtechnologies.com/products/sbr.htm
*http://sites.google.com/site/wenchiehlee1020/Papers/ComplexFB_AES_121paper144.pdf
*http://www.hindawi.com/journals/asmp/2009/468971.html
*http://focus.ti.com/lit/ml/sprp522/sprp522.pdf

Latest revision as of 12:36, 26 March 2011

Spectral Band Replication or SBR is a lossy scheme developed by Coding Technologies (now part of Dolby) to efficiently reconstruct the high frequency data of an audio signal from the low frequency data. SBR is paired with a lossy audio codec (known as the core coder) at running half the target sample rate. Each audio frame is decoded like normal. The decoded samples are then pulled into the SBR decoder along with encoded SBR side information and the high frequency data is generated and recombined with the low frequency data. The full spectral data is then outputted at the target sample rate. Thanks to the standardization of of HE-AAC, technical details are available from ISO.

SBR QMF filterbank

At the heart of SBR lies a 32 analysis filterbank and an 64 band synthesis filterbank. The SBR QMF is also used by Parametric Stereo and MPEG Surround.

A description of the SBR filterbank can be found in 26.404.

Codecs Using SBR

Links