limiter/dithering

[DuX]

hmmm limiting is not the same as compressing , although they have a few things in common: limiting is compressing with infinite ratio and knee set to hard, also compression influences the signal in whole, while limiting influences peaks [transients].

Cheers.

[IIRs]

Bk to limiting - is it redundant then to use compression (at least on vocals) & then to use final mix down limiting? For that matter, w my limited engineering experience, i seem to get a hotter signal on vocals w normalizing than compression?? Is that OK, or does anyone ever use both? Thanks

Don't ever normalise anything ever.

At best its pointless, at worst it permanently and irreversibly degrades your audio.

Normalising does exactly the same as turning the channel fader up, except it is destructive.

Compression otoh will reduce the dynamic range of a signal: if you compress a signal with no make-up gain and then normalise you will end up with a louder signal than if you simply normalise the original.. but normalising is still pointless.

With a modern 32-bit app the only place you need to worry about clipping is the mix buss, so compress a signal as much as you need, apply as much make-up gain as you need and off you go.. if you plan to use a limiter on the final mix there is no point normalising that either.. basically don't normalise.

The only time I can imagine it being useful is if you have a signal recorded so low that your host's mixer can't apply enough gain (wouldn't be a problem in Tracktion of course ).. but even in this case, it would be better still to record the signal at a sensible level to start with.

[audiobot202]

Bk to limiting - is it redundant then to use compression (at least on vocals) & then to use final mix down limiting? For that matter, w my limited engineering experience, i seem to get a hotter signal on vocals w normalizing than compression?? Is that OK, or does anyone ever use both? Thanks

Don't ever normalise anything ever.

At best its pointless, at worst it permanently and irreversibly degrades your audio.

Normalising does exactly the same as turning the channel fader up, except it is destructive.

This is not my understanding of normalisation at all. Normalisation is not like compression. It simply raises the overall volume level of your waveform by rasing the peak signal to whatever you set the normalisation to. So if you shoose 0db, it will raise the biggest peak in your waveform to that level, thereby rasing the overall volume. Here's a definition:

" The act of normalising audio data is to change its peak volume up or down to a new level whilst still preserving the relative volumes within the track. Normalising samples to the same level will help when it comes to mixing the samples (as they all start out at about the same volume). Normalisation levels are measured in dBs from the top of the track (i.e. -3dB) or as a percentage of total volume."

Many audiophiles would actually argue that compression should be thrown right out of the windo and all you need is some EQ and then a final normalisation. Compression does much much more to destroy dynamic range and hearing than any normalisation process. Think about this: When you walk outside and listen to the birds, are the bird chirps compressed or are they full of natural dynamic range?
Yes, it's like turning up the volume, but as long as your original signal is actually decent, that process will likely only raise it by a few db at best.
Where it is not good is if you normalise all your tracks individually to say 100% and then continue the mix...that leaves you with virtuall no headroom and then clipping may occur, but I don't see how it's that problematic if used once at the end to raise the signal level without clipping. If your normalised signals are ever clipping, I'd suggest you had clipped signals originally !
I'm not a big advocate of overcompression. It can be useful to cut through a mix sometimes, but I prefer a 'less is more' approach. But then again, I don't make dance music at all...so it may not ever apply to me !

[kylen]

Normalising does exactly the same as turning the channel fader up, except it is destructive.

Meaning that most apps have to do a render when using the normalize effect, as opposed to when moving a fader which does not require a render and thus is non-destructive.

You're not saying that normalization hurts the audio and samples in any way shape or form any worse than any other DSP operation.

Just wanted to plug that in before someone misunderstands and spreads a forest fire across the internet.

[DuX]

Well, compression of individual tracks like vocals, drums and bass is a necessity, but compressing the whole mix is not.

[audiobot202]

Normalising does exactly the same as turning the channel fader up, except it is destructive.

Meaning that most apps have to do a render when using the normalize effect, as opposed to when moving a fader which does not require a render and thus is non-destructive.

You're not saying that normalization hurts the audio and samples in any way shape or form any worse than any other DSP operation.

Just wanted to plug that in before someone misunderstands and spreads a forect fire across the internet.

I don't think you should be normalising on the mixdown n any case. If done at all, it really should be left to a decent audio editor as the last thing and only then if it sounds fine. Mixes should really be left as virgin as possible. By all means, get the volumes right within the mi to create dynamism when listening, but apart from light compression on some individual tracks or some EQ...I would not hurt the original mix too much at all. Leave all that for your audio editor.

[kylen]

I don't think you should be normalising on the mixdown n any case.

I'm just responding to the idea that normalizing would 'hurt' the audio any more than any other DSP operation in my earlier post. I don't believe that's the case. There are plenty of DSP developers here that might have an idea about this too.

As far as when and where to use a tool - that's not for me to say except that I use effects, faders, and volume envelopes to adjust track levels in mixes and during pre-mastering. I generally use normalization when I'm testing a new plugin, app or during some other maintenance activity.

[audiobot202]

I'd like to know how it might hurt the audio..normalisation that is. 'Platinumears' suggested this would be the case.
I really don't see how the actual process can, unless you're working with overdriven, clipped or overcompressed audio in the first place. A signal does not have to be at 0db to actually sound good Many times I've analsyed a commercial track in Cool Edit and noted the squaring off of peaks - digital clipping - and the massvie volume level across the whole track which suggests overcompression. That's not a good thing....but as far as I know, certain types of music, classical for instance, do not ever use compression for recording. By its very nature, compression destroys dynamics. It may be useful for some recordings, like rock or dance, but it's not a general rule of thumb by any means.
Some producers simply EQ and normalise a few db to complete a mix and as long as the original song is mixed well, has good resolution and all the parts sit well with each other, the ear will perceive this as a good sounding and natural mix.
If you raise the volume of peaks relative to all other parts of the signal, I'm not sure how it could be considered to be destroying the audio or causing it to be overdriven or clipping??? As was suggested earlier in the thread.

[kylen]

...but as far as I know, certain types of music, classical for instance, do not ever use compression for recording.

Here's something you might want to consider too - or discuss concerning compression. When more of the room and air is included in with the recorded instrument(s) then maybe less compression might be reguired - it depends. The size of the room, objects in the room as well as the air and distance to the mic all tend to shape the eq & dynamics of the sound and balance it out naturally. The closer you move a mic to a sound source then the more you pic up the detail & definition of the instruments' own natural resonances, timbre and dynamics - thus the need for additional compression in a lot of cases. But that's not necessarily a bad thing if used tastefully. So now we begin to get into the subjective and one man's overcompression is another man's pumpin, rockin, happenin sound.

So the fact that classical recording may not use a lot of compressors doesn't mean there isn't a lot of compression and balancing happening in the room. If we were to stand 4 feet away from a pumpin, chuggin, blastin string section and mic it I'm guessing it would sound a lot different than if we mic'd it further out in the room where the sound balances out a bit better.

[rpc9943]

he's saying destructive as a term. You can't undo it after it's saved. It's like using adobe audition in multitrack mode, its non-destructive, whereas in edit mode any effects are destructive.

RonC

[audiobot202]

he's saying destructive as a term. You can't undo it after it's saved. It's like using adobe audition in multitrack mode, its non-destructive, whereas in edit mode any effects are destructive.

RonC

The quote goes like this:
"Don't ever normalise anything ever.

At best its pointless, at worst it permanently and irreversibly degrades your audio.
"

I understand editing destructively, but this sounds more like it should never be done.

[audiobot202]

...but as far as I know, certain types of music, classical for instance, do not ever use compression for recording.

Here's something you might want to consider too - or discuss concerning compression. When more of the room and air is included in with the recorded instrument(s) then maybe less compression might be reguired - it depends...

Very true. I was thinking more in terms of DSP compression post recording, rather than naturally occuring or room dynamics.

[kylen]

he's saying destructive as a term. You can't undo it after it's saved. It's like using adobe audition in multitrack mode, its non-destructive, whereas in edit mode any effects are destructive.
RonC

Yep - that's what we're talking about - well put! Just trying to clarify a couple of things...

[audiobot202]

Don't ever normalise anything ever.

No. That's what I'm talking about actually. I'm asking why? I'm not discussing levels of undo and redo.

[IIRs]

"At best its pointless" ie: your audio file is 32-bit and the normalisation process is exactly the same as raising the gain on the mixer channel, except you can't change your mind about it.

"at worst it permanently and irreversibly degrades your audio." ie: your audio file is 16-bit, and normalisation introduces an extra process followed by an unnecessary truncation back to 16-bit. This will degrade the audio quality, even if its not always apparent with all material, and this kind of degradation is cumulative so should be avoided at all stages whenever possible.

In a 32-bit app it doesn't matter if the channel "clips" as you have bags of headroom, so setting the highest peak of any one channel to 0dBFS is pointless.

If you plan to release an "audiophile" version of your mix with no mastering processing, then by all means normalise the mix (to just under 0dBFS) before dithering to 16-bit.. but if you plan to use a limiter at the end of a mastering chain, normalising is again pointless on the 32-bit version, and damaging to a 16-bit one.