Is it peak normalization or RMS normalization? From what I have seen, it does not seem like peak normalization. It's more like a form of RMS in that it averages RMS energy over entire file, unlike Replaygain which averages over 50ms blocks. Does any body know what type of algorithm does Soundcheck use?

Anyone??

My suspicion is that its less likely to be peak normalisation. I usually have it on though I read on ipodlounge that turning it on will drain your battery more. (I doubt it is that much to be easily measurable)

You really shouldn't believe everything you read. Especially not anything you read on iPodLounge.

Turning on SoundCheck will, in fact, make your battery last longer in the vast majority of cases. Although the difference will most likely not be truly measurable. The reason for this is simple: SoundCheck usually has the effect of reducing the volume on most tracks. It doesn't have a large amount of wiggle room to raise the volume, but it has a huge amount of room to lower it, and this is what it generally does. Lower volume = less power used = more battery life. Simple.

Anybody telling you that SoundCheck drains the battery is on crack.

I seem to remember reading that SoundCheck is based on RMS. I can't remember where, and I can't say if what I read was true.

It could definitely lead to greater battery usage due to having to "crank" the volume higher to hear reduced tracks.

Also, I believe soundcheck is very closely related to the replaygain algorithim(s).

It's RMS but seems to take fewer samples for analysis (it's very fast). You can check the value that replaygain and soundcheck give, and they're usually fairly close.

Assume you have two versions of a song, identical except that one is mastered louder. You play them so that they both sound equally loud. In other words, the volume control is set higher on the softer one than the louder one.

Will they use different amouns of battery power?

Generally, the one mastered "hotter" will use more battery because of a higher bitrate (VBR only).

I plotted a bunch of values for songs generated from the two algorythms a while back when Areo was working on foo_pod. There generally wasn't a huge difference, though there were strange samples where the two would be 10+ dB apart (generally on very quite material). One of the songs on Perfect Circle's second disk was an extreme example of that.



Agreed. That place is filled with misinformation.



No. However thats not really relevent. What Otto is saying is that in generaly RG/SC almost always make songs quiter, so if you enable it, the result will be a (very slightly) reduced amount of power dissipated by the amplifer.

I dunno. People fiddling with the volume more often might overcompensate and thus play their music louder.

In any case, SoundCheck itself won't cause the battery to drain more.

This makes sense to me. I was thinking that it averaged over an entire file, not just longer blocks. I have updated by blog with this new info.
http://manishbansal.blogspot.com/2005/11/h...distortion.html