Scanning a mix

In nTrack we can scan (normalize/scan) a wavefile or some part of it.

Is it also possible (without mixing down) to scan a mix ?

/Goran Sweden

“Edit–>Normalise” will normalise a track. To my knowledge you can’t normalise a mix unless you re-import it into n-track. I tend to do that in a sound editor though when I top & tail the file etc.

Is this feature available in other audio sotwares.
It would be very easy to implement in nTrack.


/Goran Sweden.

<!–QuoteBegin>

Quote
It would be very easy to implement in nTrack.


I’m not sure I agree. n-track would have to do all the work necessary to create the mix file before it could normalise it… and then what would it be normalising? Something that doesn’t exist. I don’t get it, or the point of it. Please explain why you would want to do this?

It would be very easy to implement in nTrack.

1.) Mixdown and keep the mixdown within memory.
(Do not save as file)

2.) Scan.

3.) That’s about it. :p

/Goran Sweden

But why would you want to do it? As soon as you change anything in your mix you’d have to do it again?

And at 16/44.1 a 5 minute song is 50Mb. I’d hate to think how big the file would be at 24/96, or even 32bit formats. Huge. Don’t want to hold that in RAM.

Some 100MB is not much.
There is no problem with this.

/Goran Sweden

I don’t know why you would want to do such a thing, but isn’t what you’re asking the same as putting compression and a master limiter on each channel so each channel has the same constant level of volume? With the no fluctuations in volume wouldn’t a listener tend to suffer burnout? Maybe I’m wrong, but I have always thought some difference in levels keeps the ears awake.

cliff
:cool:

I just want to know the headroom of the mix.

/Goran Sweden

If your loudest channel is peaking at 100% and the rest of your mix is below that, your headroom is still 0. Or do you want all the channels at the same level?

cliff
:cool:

Goran makes perfect sense. Instead of having to listen to the whole mix (with “Options -> Playback meter anticipatest output” checked), and remembering to note the peak level BEFORE hitting stop or the song ending (with auto-stop) [GRRRR - that really pisses me off that stop clears the peak level! Ok back to the thread.]

Well, instead of all that, N could just run the offline mixdown, NOT save the data at all, and just keep the peak level.

What I do is mixdown and save to 32 bits, import the track, and then normalize that. Then I apply effects on that track (in the same song file for a single, otherwise in a .sng file for the CD with all the tracks) to master it.

When using 32-bit tracks, it doesn’t matter much if you’re peaking out at the perfect level, just so long as most of the mix is in the green zone and not in the weeds. (Because you’re LISTENING to it, and anything more than a little clipping, or a signal in the weeds, won’t sound good through your soundcard.)

Really, guys: Goran is 100% on target here. It’s a convenience feature and not necessary because you can work around it, but certainly would be useful and handy. I’d use it myself, frequently.

What do you guys do? Ignore the master level? Or watch it like a hawk? That shouldn’t be necessary, but other than the kind of workarounds I posted above, it’s what you have to do.

What do you do with the results? Simple: adjust the master fader to compensate. If you’re peaking at +3.2 dB, back off by 3.3 dB on the master fader. (Why 3.3? A pinch extra to stay safe, and n-Track isn’t perfect at rounding dB values anyway, I’ve noticed. Hmm, maybe 3.4 dB.)

Cheers
Jeff

<!–QuoteBegin>

Quote
What do you guys do? Ignore the master level? Or watch it like a hawk? That shouldn’t be necessary, but other than the kind of workarounds I posted above, it’s what you have to do.


With 32 bit mix downs, I never cared really, so long as it was peaking in the -3ish range or less, I don’t care really.

Well, sorry then. I am by no stretch of the imagination a master mixer. However, I don’t normalize my mixes because my tracks tend to vary from soft to loud and back again - which is really what I want…I think.
Now that my lack of expertise on the matter is evident, I guess I should back out of this conversation.

sorry Goran

cliff
:cool:

Four words: Kjaerhaus Audio Master Limiter – push her 'til she squeals! :D

Seriously. I can get my home brews just as hot as any commercial CD if I want by using the Master Limiter on n’s Master Insert and then tweaking in Adobe Audition using EQ for polish, maybe a little multi-band comp if needed and then Hard Limit for LOUDNESS! Audition is a great proggie BTW, worth every penny IMO. I got V1.5 CHEAP a while back and am looking to upgrade to V2.0 soon.

Stupid meters… your eyes can’t hear… :D

I pay attention to the REC Vu’s because I don’t wanna record a bunch clipped garbage but by recording at 24 bits even that is not much of a problem. Playing back tracks at mixdown time, I keep the meters in the greenish/yellowish area and then slap the KA Master Limiter on there and tweak to taste. Having said that… I HATE over-sqashed CD’s. The ones that are so “in your face” get on my nerves in about a minute and a half.

D – Chairman of Dynamics Saving Associated Cooperative (CoDSAC for short! :D :D) LOL!

I have actually thought about emailing in a feature request for this. The way I would like to see it done is to have an option of displaying timelines of any signal point in the system with the background color (at each point of the timeline) reflecting the meter colors and/or have a line like the volume envelopes that indicates the peak levels. By also alowing you to display a timeline of the mixdown (or groups or input to a compressor or…) you could get an overview that would show what regions of the mix might require attention. You could also display this information on the individual tracks, preferably simultaneously with the various envelope traces. In this way you could selectively make adjustments and see where you are in terms of clipping and try to identify which channels are contributing the most to the the crest factor. This in turn might suggest ways to adjust the mix to make things louder without excess use of compression.

The problem with using the scans for this purpose is that they only find the highest peak, if you knock that peak down you have to rescan for the next highest peak ad infinitum. With the whole mixdown visible as well as the individual tracks you could tell at a glance all the regions that needed attention and the effectiveness of any measures you take to solve the problem. While a change such as altering the compression on one channel would require the whole mix to be re-scanned, changes such as volume envelopes would only require localized recalculation and could be made instantly. With modern computers this whole deal would not take a huge amount of time and would be extremely useful for maximizing loudness without compression or limiting as well as for optimizing the performance of a compressor/limiter. Being able to disable the feature would allow working more quickly when you are dealing with other issues.

For me this would be very useful because I work with generally acoustic music and overall compression is very bad sonically. A compressor with a single band will squash the whole mix in response to one peak. It would be much better to bring that one peak dowm on the individual channel, perhaps with an envelope adjustment so the level of the other tracks stays constant. Multi-band compression on the master channel may be better but basically rebalances the EQ every time compression is required in one band. If it is quick you might not notice it.

We could start a normalization thread (although the archives probably already contain a lot of discussion on this issue) but I will make at least one comment. Normalization does not equalize levels, even if rms normalization is used. In the case of peak normalization, there is no reliable relationship between peak level and loudness. If a cymbal hit coincides with the peak pressure from a bass note and several others you can get a really high peak. If it coincides with a pressure minima, the same hit may not show as a peak at all. Rms normalization is a bit better since it responds to a long-term average but if you have a song which starts quietly and gets louder it will have a very different rms level than a song which is more consistent over time.

There is no substitute for listening. I typically adjust my song-to-song levels by listening to the vocals and setting the master gain so that they represent the levels you would get from a live performance. A balad should not be the same level as a “belter”.

Meters do not know about music, they only know about physical properties. Once I have the relative (song-to-song) subjective levels correct I look to see what I can do to increase the overall level. It might be as simple as finding that one really high peak and selectively using an envelope to knock it dowm 3 dB so that all the tracks can be increased by that amount. If only one or two high-hat notes need attenuation it would be silly to reduce the level of all the instruments in the mix to solve that problem (ie. use compression) but if the peaks are all over the place it may be that compression is the best answer. A timeline for the mix and “stationary” level indication would be very useful for determining what to do to best preserve the performance while increasing the level.

Jim

Well, so much for the “Stability” votes! :laugh:



(sorry!)

On this: If this is relevant, I found that using envelopes (volume) for taming stray peaks in individual tracks works much better than compression. No residual effects on the rest of the mix, for all intensive purposes, anyway.

Quote (g8torcliff @ Mar. 14 2006,16:21)
Well, sorry then. I am by no stretch of the imagination a master mixer. However, I don’t normalize my mixes because my tracks tend to vary from soft to loud and back again - which is really what I want…I think.
Now that my lack of expertise on the matter is evident, I guess I should back out of this conversation.

sorry Goran

cliff
:cool:

Cliff, you’re confusing normalization with compression.

Normalization just shifts the whole signal louder to peak at zero, and does not affect the dynamic range. Compression reduces the difference between loud and soft. And, while you do indeed want the track to vary from loud to soft, this does not mean that you shouldn’t use compression. That seems like a contradiction, but it’s more like a warning not to overuse the tool.

Mixes that don’t use any compression at all usually sound quite weak and amateurish. Unless they’re done to perfection, in which case the result is impressive – but that takes perfect performances as well as perfect engineering, in addition to careful attention to arrangement and other matters.

You’ll hear lots of audiophiles rage against compression, and they’re quite correct – however, what they’re really raging about is overuse of compression, or just bad judgement about dynamics. To revise an old slogan: “Compressors don’t ruin mixes, engineers do.” :;):

Jimbob makes some good points, but there are some significant complexities to the feature suggested. I touch on a couple in the wiki page that Jim created.

Jeff,

I agree with many of your points about the feature and would be happy to have the feature use a “refresh button” to initiate the updates.

I am not sure about your assertion that the word length precludes clipping. In part that would depend on where the “binary point” was and how things were scaled to begin with. You may be completely correct but it would be possible to design a system with any number of digits above or below the binary point which might limit the headroom in order to get better performance with attenuated signals. Without knowing the details I certainly couldn’t say. I do know that compressors can and do clip if the input is too high and I suspect that other plug-ins can as well. This is because they ultimately have to fit into 16-bits and when all is said and done would have to either normalize everything to full-scale or truncate above a certain point. Poorly designed equalizers (for instance) can also have much larger numbers internally than at either input or output which can result in internal clipping that is not identified by either input or output levels.

I have a suggestion to avoid having the peak levels disappear on the meters when the playback stops. Disable the “stop playback at end of song” option. While you might forget to press stop, there are no consequences to this (at least on my system) as long as you don’t try to use your soundcard with another application while playback is on-going. I’m not sure what happens even then, maybe I’ll give it a try tonight. Anyway when I am working on a section (I do not have the view follow the cursor which I view as a bad idea most of the time - impossible for envelope work) and the playback continues past the end of the song, it just goes silent. If I want to listen to my section I just click on the timeline and it plays where I click (which is slightly more convenient than pressing play and then clicking on the timeline). In your case the meters will hold their values until you restart playback (either by clicking on the timeline or pressing stop then play) so you don’t have to worry about spacing out and forgetting to look at the right time. Hope this helps.

Jim