Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Weird SBR limitation/question (Read 942 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Weird SBR limitation/question

I've noticed through some tests something that I couldn't express with words for a long time.

Every now and then I would stumble across content via DAB+ or internet radio or some streaming service that would use somekind of codec at low bitrate with some form/implementation of SBR. I know what to expect (in an audible way) and I have a rough idea of how spectral band replication works but there was always that "odd" feeling every now and then when I would hear content that I am familiar with and it felt like I was hearing a different version of what I'm expecting to hear.
I was finally able to put a needle on what that thing was.

It seems that content that have it's foundamental frequency or what humans perceive as it's most important frequencies, are out of the "baseband" sampled data are going to be gone from what you will eventually hear. For example Glockenspiels, bells or anything that it's foundamental frequency is out of what is going to be in the main part of the codec, is completely invisible to the SBR implementation.

Ofcourse, this tends to be the case with relatively low bitrate content rather than higher, but my question is this.

How do different implementations of SBR work on different codecs with issues like this?
Are there any current or developed/proposed implementations that are able to sort-of alleviate the issue?

Re: Weird SBR limitation/question

Reply #1 –
If I understand you properly: SBR has a very coarse way of reproducing high-frequency harmonics (by copying "up" lower-frequency harmonics or even the fundamental tone of a signal). This may not work very well on inharmonic signals such as Glockenspiel or triangles, resulting in some audio content to be missing some higher-frequency tones after compression with a codec using some flavor of SBR (HE-AAC, e-AC-3 and AC-4).

There are two techniques to alleviate this issue: one is a "missing harmonics" tool in some (but not all) SBR implementations - IIRC, this may not be supported in DAB+ e.g. - the other is a more recently devleloped high-frequency reconstruction technique called intelligent gap filling (IGF), which is only available in the latest audio codecs like 3GPP EVS and MPEG-H Audio - it cannot be activated ino legacy codecs like HE-AAC.

Chris
If I don't reply to your reply, it means I agree with you.