Topic: Does a tape recorded at -1dB ultimately sound 'louder' than one at -12dB??

[yates7592]

I routinely record my shows with peaks hitting the -12dB mark or thereabouts in 24 bits. I do this because I stealth and I can't fiddle with my levels, I like to set and forget. I then normalise to -0.1dB in post. Of course they sound fairly quiet unless some plug-in work is carried out. If I were to run my levels much hotter, say peaking -1dB (difficult I know), will that recording always sound 'louder' (in RMS terms) than the normalised -12dB pull, and if so, why?

[OOK]

My answer is yes.  It will sound louder becuase it is louder, however you run the risks of overs/brickwaling if you run that hot.  What I am surmising from your question is doe sit sound better and my answer is no.  It is just louder. 

In the 24 bit realm you are better off running conservatively and bringing up the levels in post like your currently doing. Unless you realy know your gear and venue then you can up it a little.  I tend to run my levels peaking around -4db for the venues I hit based upon my experience.  Venues that I am not familar with I back it down a little.   

hope this helps.  OOK

[yates7592]

Thanks OOK for your reply. I think what I am asking is:

1. If both recordings are made one peaking at -1dB and one at -12dB, then BOTH normailsed to -0.1dB, both recordings will have the same peak level (-0.1dB). But the recording made at -1dB will probably sound 'louder' in RMS terms, rather than on the peak meter. Is this correct? It's as though the 24 bit recording made at low levels is 'missing' something, like all those bits and bytes aren't utilised to their full potential.

2. If so, why does the RMS value apparently increase for the hotter level pull when both recordings ultimately have the same peak level?

[JasonR]

Within reason, they will sound the same.

The one recorded at -12dB will have a higher noise floor, but with 24-bit and recordings of amplified music (not crickets and mating frogs), the noise floor should be relatively irrelevant.

I suppose the one recorded at -12dB may also have *slightly* less dynamic range, so upon increasing both to the same peak level, the -12dB one will have a higher perceived volume as it's effectively compressed slightly by the difference in dynamic range.  Again - if we're talking 24-bit, -1dB vs -12dB, and recording amplified music, it's all pretty irrelevant.

Perhaps the difference could be more significant with a sub-par recorder.

[page]

Jason and Jon beat me to it.

[twatts (pants are so over-rated...)]

[yates7592]

Yeah, I know how to use a volume knob but I was not asking "how do I make my recordings louder....."

[twatts (pants are so over-rated...)]

Yeah, I know how to use a volume knob but I was not asking "how do I make my recordings louder....."

Yeah, but this one goes up to 11...

Terry

[datbrad]

My take is that for stealth, or any situation where you can't get to the deck even if you wanted to, not having to worry about overloading is the primary concern. That being said, to my ears the answer on "loudness" is the opposite of the OP's question.

When I make a recording with peaks very ocassionally hitting zero, compared to a recording that I recorded down to peaks in the -6 to -4 range, and boost them in post (not normalized - just amplified), the file sounds louder, but not in a good way. Typically, I find them to be a tad harsh and "cranked up" sounding for lack of a better term.

My theory follows the notion that when more bit depth is used for the master recording, even though true 24bit dynamic range is never supplied by the source, or the analog path in front of the A/D, it just seems to sound better in my experience to roll hot at the show, and not have to do anything else, rather than set levels modestly and have to jack them up later with software. Sure, it's more work at the show, but it's a waste of high resolution PCM not to maximize it's capabilities. Now, I will say that if you are going to leave the rig for a large portion of the night, which I've done plenty when the situation called for it, you do have to roll lower.

[swordfish]

My take is that for stealth, or any situation where you can't get to the deck even if you wanted to, not having to worry about overloading is the primary concern. That being said, to my ears the answer on "loudness" is the opposite of the OP's question.

When I make a recording with peaks very ocassionally hitting zero, compared to a recording that I recorded down to peaks in the -6 to -4 range, and boost them in post (not normalized - just amplified), the file sounds louder, but not in a good way. Typically, I find them to be a tad harsh and "cranked up" sounding for lack of a better term.


2nd that

I also always have the feeling the the boosted recordings sound different as the ones which are recorded at the limit..

[F.O.Bean]

Friday night I recorded umpires McGee. First set I was running hot and peaked at about-4db. Thought it was too hot so second set I lowered my levels and peaked around -10 to -12db. Normalized both in wave lab 6 and both sets sound exactly the same. So I wouldn't worry about peaking at like -3db and just roll conservatively atthe show. Much easier than worrying about peaks overloading and fine tuning during the show. I run conservatively at the show and normalize to 0db in post and am quite happy with the end result

[Gutbucket]

Back on-line today, I'll add a bit here..

Remember that loudness and level are related but not identical.

Level is easily measureable, loudness is perceptual and often easier to hear than measure accurately.  Obviously if you increase level, loudness also increases, yet two recordings with identical peak levels can produce quite different perceptions of loudness. 

Some important terms-
Peak level is an instantaneous measure of the highest level achieved.
RMS level (Root Mean Squared) is an average measure of level over time, which better approximates the human perception of loudness.
VU meters (Voltage Unit) are similar to RMS metering as they also integrate level over time, hiding true peaks.
Crest factor is the ratio between RMS and peak levels.  It is a large part of what we wrestle with in setting recording levels appropriately.  The crest factor varies significantly depending on the type of music, the instrumentation, the performance venue, and other variables.  Amplified music produced through PA with limiters protecting the amps and probably other compression in the mix, that has a relatively constant loudness level during performance will have a relatively low crest factor.  Hyper compressed commercial recordings which are casualties of the 'loudness war' will have a very low crest factor.  Recordings of things with large dynamic swings can have high crest factors and need more headroom.  Imagine a close recording made of an unaccompanied wood block player in a quiet room.  Very short, high-level peaks with a bit of reverberant decay down to the room's noise floor between 'hits' makes for a very high crest factor.

Increased loudness for the same peak level can happen in various ways, most commonly by compressing, limiting or clipping the peaks of the music and adding makeup gain to bring the resulting signal back up to what would have been the original peak level.  The average level will then be higher relative to the peak level than it was previously, and the crest factor of the resulting recording (though not of the source being recorded) will have been reduced.  Compression, limiting and/or clipping can happen while recording or be imposed upon it later.. either intentionally or not.


One thing is certain.  No one here is ‘loosing resolution’ by recording at lower levels and amplifying the resulting file later unless doing so brings the noise floor up enough to become audible in a bad sounding way.