Amplitude Question for Audiophiles Smarter than Me

As you have no doubt noticed, I’ve been posting quite a bit of Surround Sound posts during the last several months. One of the things I do in the process of getting the Surround Sound .flac files ready for posting is to run them through an amplitude analysis (using Adobe Audition CC 2015) to determine the file’s maximum peak level, and then using the result of that analysis to turn up the .flac files’ volume to their collective maximum possible level without going over the "0 dB" threshold, so as to give everyone the greatest possible output while they are listening. So far I haven’t gotten any complaints about how I’m doing it, so I assume it’s been successful up to this point.

One thing I’ve noticed, however, is that, without one single exception, every extracted surround sound file I’ve worked with always requires an amplitude increase of some sort. I have never found an SACD (or Audio DVD) where the Surround Sound files were already at "peak", so that I could simply post them "as is". This is not true of 2-channel files, especially with standard 44.1 kHz/16-bit redbook CD audio, as they are are occasionally already mastered at "peak" and require no changes before posting. So why are Surround Sound files different?

The reason for this post here is because I’m starting to wonder if, perhaps, there might be a reason for surround sound files always being below peak, which I might be unaware of.

Is it possible that, by turning up the volume level to the maximum level short of 0 dB, I might be causing a problem that I’m not aware of?

Any comments or suggestions you can offer are welcome.

(It would be especially helpful, also, to hear from anybody who has downloaded some of my Surround Sound files and listened to them, who could give me some feedback on how they sound, if the way I’m preparing them seems to work OK on your end.)

(And moderators, if I have posted this in the wrong place, then please feel free to move it to a more appropriate location with my apologies.)

If you downmix a surround track to 2-channel when the individual channels are already at peak, it will suffer a lot of clipping and be ruined.

It would only be suitable for playback in surround but would suffer if anyone tried to play it back in stereo or if the track were downmixed using a software and saved. Lowering the gain would be necessary before downmixing and saving it. In the end surround tracks being at a lower overall volume is always for the best.

I’ve made surround files where I’ve increased the gain to be near peak level, but I’ve done that knowing they’re only going to be played back in surround and not in any downmixed form, unless I’ve mixed the center channel into the front stereo first….I don’t actually use a center channel speaker in my setup.

I think I see what you’re saying, because if multiple channels at peak are being squeezed by the listener’s system through only two channels, then overloading would result and a lot of clipping would occur.

So then, since I also provide 2-channel options alongside the surround tracks, I need to caution people to only take the surround option if they intend to play it in a truly Surround environment?

I believe the mastering engineers deliberately set it to peak around -3dB or so. 0dB is the maximum possible, so they set it to -3dB (it has been lower on older recordings, but I believe the current standard is -3dB). I believe the reason is to allow for headroom for processing as well as level mismatch between equipment, especially between processors and amps. And to give processing headroom – when you’re applying filters and other things to the signal, you may encounter excursions above and below, and using -3dB means the filter has the digital headroom to go 3dB higher without clipping the signal. It’s the same reason why video generally doesn’t go from 0-255 but 16-235 or so – so any video processing has room to make excursions into the clamped area but still retain full fidelity of the processing instead of saturation.

LFE is set to -10dB peak on recordings – the surround processor boosts LFE +10dB during processing to bring it back up to normal levels. I don’t know why this is.

If I may, it is my undestanding that the due to the signal being digital, ie not continuos but sampled in time, upon reconstruction of the analog signal the result may be bigger (in amplitude) than the adjacent in between samples. The phenomena has been described at "true peak" as compared to peak. You may google it for a better explanation. This is why all recording are not "normalized" to 0dBFS but a lower value, perhaps -3dB as Worf is suggesting. Now beyond that most SACD recording are being now "normalized" to -23dBFS. They can do this (very low level) because the recording is usually in 24bits of depth rather than the most used CD format with 16bits so there is actually no sound degradation. Now they do this because when you reproduce this on a "calibrated" audio system it will provide more or less 85dB SPL wich is considered the sound level your ears should be receiving from the speaker for "critical" listening.

Cheers

This is very interesting. I’m still learning (obviously), so I will definitely Google "true peak" and read up on it. Thanks! https://dl.dropboxusercontent.com/u/42753709/Photobucket/yo.gif