I have no technical understanding over how OGG Vorbis works, so here's some random thought.

HE-AAC sounded great because of SBR.
Try encode 48kbps AAC using LC profile, the quality sounded far worse than OGG Vorbis at 48kbps (AoTUV5).

I'm a Vorbis fan, I just feel a bit dissapointed the HE-AAC clearly outperforms OGG Vorbis at 32kbps/48kbps.



Whoops... Sorry guys.
Somebody has discussed this before.

http://www.hydrogenaudio.org/forums/index....showtopic=43277

Despite the quality enhancements SBR brings with itself there's no real need for it to be used in conjunction with a format like Vorbis, simply because this codec already delivers good performance in the low-bitrate regions. SBR works best with AAC-bitrates of 80 kbps and fewer, the ones usually used for portable players, leading to impressive quality. But the fact that it's only useful for portable-bitrates is also its major drawback, since it needs a lot of additional processing power and therefore has a noticeable impact on battery life. For this reason I clearly prefer Vorbis at ~80 kbps to HE-AAC at 48 or 64 kbps, i.e. even if SBR/PlusV was available for Vorbis, I'm pretty sure that I wouldn't use it anyway.

aoTuV b5 makes -q0 (~64kbps) sound pretty decent, to be honest, especially with a raised lowpass (the default appears to be too aggressive). Of course, lower than this, HE-AAC (without PS) does sound quite a bit better.

It's been discussed a dozen time, plus being a conosseur of fine tastes SBR doesn't suite me very well. Vorbis uses a more natural approach called Noise Normalization. It's more eloquent and it's not patented. If it can be tweaked correctly it can give impressive results, which what I think Aoyumi is always consistently attempting to do.

Can someone tell me if the extreme variability of Ogg Vorbis VBR poses problems for use in streaming, as on the internet, satellite or cellular phone networks?

I use Oddcast with Lancer DLLs at -q-1 (~48 kbps) and don't have any problems; the bitrate never jumps higher than 51 kbps. Oddcast seems to use ABR, but I can't confirm that.

Honestly I do not see the connection between SBR and noise normalisation.

There is no connection with the way it works, but it's not patented and it's easier way of redistributing noise energy. "Similiar technology" at least the way I percieve it would work according to this. Even if it's not exactly the same.

^ Noticing Gabriel and HotshotGG above:
So, is there any project to create an open source tweak system to enhance hi-fi--ness at rates under 50 kbps? My understanding is that incorporating SBR and PNS in Ogg Vorbis is possible, but would require royalty payments. And I think Thomson has patents to wall off some tricks in its MP3Pro...
So, what alternatives can keep open source codecs competitive?

PlusV is open-source technology similiar to SBR, but I don't think xiph developers feel it's needed. NN is similiar to PNS. Considering what a lot of companies are doing there isn't a whole lot that can be done on open source front and considering a lot of latest industry standards are heavily patented. It doesn't give a whole lot of room for innovation, but then again is it really needed?

Here is Garf's opinion
http://www.hydrogenaudio.org/forums/index....c=34075&hl=

If you want to know how vorbis theoretically sounds with plusv...
http://forum.doom9.org/showthread.php?p=817234#post817234

Noise normalisation means that in a given frequency band you are trying to produce the same sound power in the quantized signal than in the the source file. It's not a coding tool, but just an encoder design choice.

SBR is parametrically recreating the higher freq part from the lower freq part.

PlusV is parametrically recreating the higher freq part, without considering the lower freq part.

PNS is recreating noise-like freqs synthetically.

(both SBR and PlusV are also using PNS)

Even if the floor representation of Vorbis is efficient, it still have to encode a lot more data than SBR or PlusV in high freq parts, and no, noise normalisation is not similar to SBR.

Note: PNS is part of AAC, AC3, WMAv2

gameplaya15143: that um, sounds... pretty lousy. Even with aoTuV b5 (and I changed your bat to use -q0 instead, because -q-1 and -q-2 sounded a LOT worse), it sounded bad.
But hey, maybe with tuning it'd be better right?

Oh, and here's some samples so you can compare for yourself. Included HE-AAC just for the sake of comparison.
30s clip of Alone Tonight by Above & Beyond.
http://download.utorrent.com/alonetonorig.wv - original
http://download.utorrent.com/alonetonplusv.wv - aoTuV b5 -q0 + PlusV; (full track was 48kbps when including plusv data)
http://download.utorrent.com/alonetonq-1.ogg - aoTuV b5 -q -1; (full track was 47kbps)
http://download.utorrent.com/alonetonhe48.aac - Winamp 5.31's CT AAC encoder, 48kbps HE-AAC (no PS).


Personally, I prefer vanilla -q-1 over PlusV, but I think it's just a matter of preference, 'cause neither sounds good. Neither are really competitive with HE-AAC.

On another note, vorbis at q-1 sort of reminds me of how MP3pro sounded.