PeakExtend ON or OFF?

[Marcin_gps]

What's your preference? I know that it depends on album's volume, but do you prefer the sound of XXHE with or without PeakExtend? I like PeakExtend off better in general, because my amp has easier job driving the speakers.

[boleary]

Does peak extension have any influence on the sound, other than volume, when playing a track that is not HDCD? Non HDCD tracks seem to sound beter here with peak extension ticked, but that just might be because of the reduced volume sounding "more resolved". Without being able to precisely measue the volume difference and add exactly 3db, IIRC , its hard to tell. My impression, guestimating the difference and trying to correct for it, is that  everything sounds better, less HF distortion at very high volume, with peak extension ticked.......Again, is this possible with a non HDCD track?

[PeterSt]

But OF COURSE !
Actually it is totally unrelated to HDCD !!

Most often tracks are clipped. Maybe more, maybe less, but as soon as the (dgital) volume is a little "over", it is often chosen for (by the producers) to "cut" the volume, instead of decreasing it. This way the track remains louder (on average), and they believe you won't hear the cut "peaks". So, at the top of those cut waves, normally a fluent peak, now here is a horizontal line coming down to a square.

What Peak Extension does, is first decrease the volume (like it should have during production) to give headroom to the virtual peaks, and next extend those peaks again so they don't have "squared" tops. Arc Prediction is needed for that, and it does it as real as possible.

There sure is a similarity with HDCD, because that too produces squared tops (to gain headroom for additional resolution by a means I'm not sure I know it ). So there too those peaks have to be restored, with the difference that it is more exactly known (??) how to reconstruct them. So ... with or with HDCD decoding, an HDCD is reconstructed for those peaks, in the latter case because Arc Prediction does it. In the former case Arc Prediction does nothing anymore, because there are no cut peaks anymore (HDCD decoding restored them).

If you now read back what you perceive from it ... this is *exactly* right up to each detail. You may highlight them yourself now.
It is also very worthwhile to see such a description from someone who didn't have a clue what is going on; It means it just works.

Haha
Peter

[boleary]

No applause, just throw the $$$$$

[boleary]

Just wondering.....since peak extension works will it go away as a checkbox? I mean, when would you ever not select it, and, if there is never a time, why a checkbox for it?

Clueless in Philadelphia 

[PeterSt]

Reasons I can think of :

- It decreases the volume by 3dB (and your 300B suffered already);
- You want the music as it is, molested and all;
- With a real HDCD you better shut it off, just in case things happen twice.

But I think in the latter case it is auto-shutoff.

if there is never a time, why a checkbox for it?

Because overhere in Holland we never like to show times in a checkbox ?

IOW, I really am clueless about what you may mean here. Haha.

[boleary]

Let me try this way: Why not put PE under the hood so it's always on?

[PeterSt]

Yea, but that I answered to already; see the first two of my three mentioned points.

Btw, the 3rd indeed doesn't count, because with HDCD it is shut off under the hood. You can't see that, but otoh HDCD does it now, so it still "looks" allright.

[manisandher]

At the dawn of 'Arc Prediction', Peter wrote:

Notice that things could really improve for SQ again if I first drop the volume a bit and *then* apply QAP. I'm working in these areas anyway, so I could apply a checkbox for that, so you (we all) can test that.
Notice that I always use attenuated digital volume in the first place, and it is really done in a good/harmless way.

Whenever I use PeakExtend, I seem to lose low-level detail. The ambience of a venue gets lost. Instruments don't quite 'breathe' fully any more. (Of course, I'm taking into account the 3dB attenuation that PE imposes and readjusting the volume control!)

Peter, are you sure that the PE attenuation is done in a good/harmless way??? Could you just check the code when you have a minute please?

Mani.

[PeterSt]

Hi Mani,

Just a few days back I finished measuring EVERYTHING ... which now reminds me of the quite some bugs I solved in XXEngine3, which I now thus must find and solve again. Oh boy, this was a lot of work. (A WARNING about SSD's ...). The volume is ok, but in AP itself I found some errors, some which *will* influence SQ. I hope I can find them again, haha.

Notice that -apart from the volume- Peak Extend is doing exactly nothing. It only gives headroom for Arc Prediction itself, and as long as the digital data is 3dB ore more under max nothing will change. But be careful, 3dB is a lot in this context, and you can bet that most of the time this headroom is consumed. You can check it with the Normalized Volume : if it doesn't allow +3dB the headroom is not there. But you will be having a hard time finding albums with 1.5dB of headroom.
At the louder passages, this means that before you were listening to raw cut waves, while with peak extend you are not. The difference will be in the freshness, and the cut waves sound the most fresh. But not good of course.

I'm not saying this *is* what you hear, but it would be a valid explanantion I think.

If I only knew where the "will influence" bug was (which form of AP) I could tell you that it may be *that* (and with lower max digital output this mattered !). But at this moment I don't know anymore. Sorry. I need some days to catch up.

Peter

PS: Saved plots of the measurements as well. All gone.