The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
Is LM1n more accurate than the rest? Dynamics Plugins
Old 10th December 2017
  #1
Gear Head
 

Is LM1n more accurate than the rest?

First, I just want to say that I compared the LM1n with some of the free meters one can download online and they all showed different results in LUFS readings and especially LRA.

Now I read that the TC dropped the price of the LM1n down to a fourth of what it used to cost.

What's going on here mates and is the price going back up, you think?

Or did TC realize that someone had better algo (maybe even giving it away for free) so they're cutting their losses, or TC thinks is so much more accurate than the rest that to prove the point they are dropping price temporarily, or they just simply caved in to the freebies pressure??

If someone could shed some light before I buy something from them , it'd be splendid.
Old 10th December 2017
  #2
Lives for gear
 

Algo for measurement is pretty much defined in loudness measurement standards. For example EBU R128 has its own freely downloadable suite of various test files with given correct results. EBU Technology & Innovation - Loudness test set

Anyone, who implement R128 meter, has to test, its compliant and results are within tolerance allowed by the standard.. There's minimum requirements for its implementations.

So no, there shouldn't be any significant discrepancies at defined loudness measurement values I, S, M and LRA, among various meters, because it's pretty tightly specified. LRA in particular https://tech.ebu.ch/docs/tech/tech3342.pdf (including algorithm description, code samples and references to the tests).

And no, I don't think, someone has "better" algo in that regard - it's either correct and compliant or not (and should be avoided).

You might have differences in UI layout, various forms of visualization, convenience features, supported formats etc. Those are valid reasons, why someone can prefer one loudness meter over another, but results should match whenever particular standard is being referenced (in presets, name, logo is being used).
Interoperability and correct values among vendors are actually one of strongest reasons for the adoption of those measurements standards.. compared to absolute international mess, which was there before.

With LM1n, I don't think, there's need to speculate about their reasons.. $29 is actually reasonable price to sell TC loudness meter plugin to already saturated market (pros already bought their meters many years ago, it's commonly bundled to better DAWs and there are few very good free/cheap alternatives).
If you like its UI, and looks practical for you compared to what you already have or can get for free.. go for it.

Michal
Old 10th December 2017
  #3
The EBU R128 specs define rather generous tolerances.

Look here: https://tech.ebu.ch/docs/tech/tech3341.pdf#page=10
The smallest tolerance you'll find in this document is +/- 0.1 LU

LRA for example mentions +/- 1LU:
https://tech.ebu.ch/docs/tech/tech3342.pdf#page=8

The True Peak detection also has pretty large tolerances of +0.2/−0.4 dB.
And the EBU R128 docs also clearly mention the general "error" to be expected in resampled TP detection, which in it's own is surprisingly large at the 4x resampling recommended by the "standard". It's beyond 0.5dB!
https://www.itu.int/dms_pubrec/itu-r...-E.pdf#page=22

Generally, EBU r128 isn't particularly precise, or well made. There's room for interpretation, and several documents contradict each other (i.e. the exact filter type used in the loudness RMS detectors, or the exact filter type and specs for the TP detection). By far the worst problem of meters is aliasing, and nobody talks about it (the refresh rate is 50-60Hz after all, 25Hz of bandwidth!). Even then, it's not a bottle-neck, it's not that relevant. Nobody noticed the heavy aliasing anyway!

Limiter 6 (there's an unlimited demo, making the meter effectively freeware) uses very high quality resampling algorithms delivering far smaller tolerances than what the standard defines. But in my opinion, it simply doesn't matter. Especially not for slow meters like those showing EBU loudness measures. I found out that women generally don't dance and sing to EBU r128 values.
Old 12th December 2017
  #4
Lives for gear
 

Hi Fabien,

I see, the spec could look like imprecise, but on the other hand I can see reasons, why they left such tolerances there and left some degree of liberty of implementation to vendors of various software and hardware devices.
The most important figure in the standard - integrated loudness, which can be used for subsequent normalization, has 0.1dB range. With some common sense and experience with practical use, metering is still pretty darn accurate, there won't be any real benefit to with making that tighter and say tolerate just magnitude lower deviations. So this allowed range doesn't look like overly generous to me.

TP part was already touched in the ITU paper. It's probably not so problematic to use some higher level oversampling and great interpolation filter in software, but there could be issue to do it computationally at some hardware meter, especially in multichannel. Of course it's just my personal explanation, it can be even less important today than it was, when 1770 and associated standards were initially released.. So when someone do dedicated HW meters today and design them from ground up for that task, it's certainly easier than at some super expensive "transitional" meters or broadcast processors from almost decade ago, which had added particular loudness standards without any hardware change just via firmware update.

I'm aware of possible errors, because years ago, I've did some homegrown offline normalizer for internal use by broadcaster for tests before wide adoption of those standards. Played with different resampling algorithms, interpolation filters with varying precision and ripple and generated "pathological" signals (basically couple of drawn samples or patterns ).. It's certainly possible to obtain some pretty significant discrepancies using those and also you might have different focus on that, when you crafted and finely tuned a TP limiter with Vlad.
But in the context of mentioned use case and mainly real world programs or music signals from our test set (those half sampling period transients are quite hard to come by, unless there are digital sync problems ), I'd say, more relaxed specification, which counts with lower specced TP meters doesn't cause much issues. And I ended with something rather quick and probably bit crude for resampling generic use, but there it was appropriate IMO there.
Also readouts usually match across different meters in practice and if there was any deviation at logged maximums, it was rather under 0.1dB, which is common resolution of TP meters.

Similar small discrepancies were with LRA figures and test materials.. Although standard allows for +/- 1 LU, commonly it's about 0.1-.2 dB among different hw/sw meters, I've tried in the length of common clips (minutes). What I rather experienced was fixed "lag" with LRA display updates from some meters.. sometimes by 1 or 2 - SL analysis blocks length. Also at start of measurement, LRA readout isn't stable with some, but it settles down after a while.
So in realtime use, it can like it's off with some dynamic material after playback start, but at full length (as judged during some Q/C), it's fine.

Anyway, it's interesting subject.. yet we haven't touched details of temporal aliasing and that filter..
And although, I don't feel, the standard or minimum reqs. tests are too loose, thank you for the post.

Michal
Old 12th December 2017
  #5
Lives for gear
 

For the OP.
I've played with demo of LM1n for a while.. It's bit stripped down (which isn't always bad), TP indicator is just momentarily lit during overs, no TP maximum is logged and there's no custom setup of loudness target or TP threshold warning, besides what you have at presets. So while you can use numbers to reach some different target (say -13LUFS), color warning won't work there.
On the other hand, it's dead simple to use, if you're doing some deliveries to compliant to included presets.
It has one big competitive advantage in Pro Tools, if you use those. There's offline AudioSuite plugin, which works faster than realtime over bounced regions. Only handful of plugins can work without actual audio output there (eg. there's Analyze button). This can be big timesaver for someone, who does that often and want to find some fixed level offset to have compliant output.

Michal
Old 20th December 2017
  #6
Gear Head
 

Michal and Fabien thanks for the brilliant comments and splendid advice. I ended up buying the meter because it was cheap and also assume everybody is using it so I better have it.

Cheers!
Topic:
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump