Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Measuring Dynamic Range (Read 42321 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Measuring Dynamic Range

Reply #25 –
Quote from: Chromatix on
The ReplayGain values are as follows:

CDDA: -0.20 dB
Badly: -5.94 dB
Very Badly: -7.56 dB
Stinky: -8.06 dB
Limburger: -5.76 dB

The Sparklemeter readings are as follows:

CDDA: 7.8
Badly: 4.6
Very Badly: 2.8
Stinky: 1.8
Limburger: 1.4

These are not comparable to the numbers posted earlier, since I've changed the calculation and display methods.  For that reason, I'll give the new versions of the numbers here:

Berlioz: 7.0
Leftfield: 11.0
RHCPs: 1.1
Manics: 1.3
Brandenburg: 4.6
Wachet Auf: 2.5

Yes, those numbers certainly seem a more accurate reflection of the perceived quality of each track. May I ask about the current calculation method, and whether the schematic dust has now settled?

Still looking (very much!) forward to trying this application for myself, as I have a few "before and after" projects I'd like to test.

    - M.

Measuring Dynamic Range

Reply #26 –
It's very simple - I doubled the contribution (still in dB) from the variance of the sparkle, subtracted 10dB, and converted the total back to a power ratio.  The theoretical minimum for this new measure is 0.1, which is achieved with a digitally silent track.  I optimised a significant bottleneck while I was at it, so now my results (including the graphs) come in very quickly, doing several albums before I can make coffee.

I've also introduced a correction factor for short tracks.  This reduces the range used for the variance calculation for tracks under 1 minute.  There is no change for longer tracks, including all the ones described so far.  The two very short tracks I have tested give much more sensible figures now, which are broadly in line with everything else on the same disc.

I've also been experimenting with measuring the long-term dynamics and the corona.  I'm not happy with those results so far, but I can see why part of pfpf was done as described.  I think I can see how to measure the dynamics usefully, but the corona will require more detailed study.

Incidentally, I think it would be interesting to see pfpf in operation.  I have a Windows machine to try it on, and I'm not afraid of a 90MB download, even though I would try to avoid inflicting it on others.

My own sparklemeter is *much* smaller than that - 10KB of source, 18KB of AMD64 binary under Linux - and it relies on a couple of reasonably small libraries (libsndfile and libpng).  On release, I would try to provide a Windows binary for those here who need it.

Measuring Dynamic Range

Reply #27 –
I can get it down to about 28MB for you. 26MB is for the LabVIEW runtime, and 2MB is the actual exe.

I'll get some webspace set up today for it and I'll let you know when it's up.

If I had a Linux box with LabVIEW installed I could give you a build for that, but sadly I don't.

Measuring Dynamic Range

Reply #28 –
Well, I've discovered something very interesting.

Those of you who follow(ed) The Orb will remember that they released an extremely poorly-received album a few years ago, titled "Cydonia".  Fortunately, they've made some more recent releases which have improved matters somewhat, but most people only know about their early-1990s heyday.

If I pull out an early album at random - say, the Adventures Beyond The Ultraworld double - and compare it to Cydonia, we get a very obvious difference in both style and Sparklemeter numbers:

Ultraworld Disc/Track | Sparkle
  • 1/1 | 5.8
  • 1/2 | 6.3
  • 1/3 | 20.5
  • 1/4 | 3.1
  • 1/5 | 6.0
  • 2/1 | 5.2
  • 2/2 | 4.3
  • 2/3 | 5.0
  • 2/4 | 8.0
  • 2/5 | 2.4

Cydonia Track | Sparkle
  • 01 | 1.8
  • 02 | 1.8
  • 03 | 2.2
  • 04 | 2.2
  • 05 | 9.7
  • 06 | 9.4
  • 07 | 1.5
  • 08 | 3.4
  • 09 | 2.8
  • 10 | 11.3
  • 11 | 0.6
  • 12 | 1.6
  • 13 | 4.4

As you can see, only one of the ten tracks on Ultraworld reads below 3.0, whereas about two-thirds of those on Cydonia do.  Tracks 5 and 6 are very unusual - 5 is mostly speech, which is expected to read about 8 or so anyway, and 6 is a very short interlude track with a sharp level transition in the middle.  On the other hand, track 11 is also very unusual in nature, resulting in the lowest reading from a non-silent track I've seen yet.

It's very obvious that Cydonia has had the compressors applied liberally, so although there is still enough headroom to prevent "clipressing", it has completely destroyed the ambience of the music.  Ambience, of course, is precisely what The Orb is known for!  The mastering engineer (and/or his manager) for that record should be shot.

It is perhaps no wonder that The Orb temporarily started their own label (Bad Orb) after that experience!

But - there is an exception.  Just one track on Cydonia retains, apparently, a low mastering level and all of it's dynamic range - and manages to sound just a little bit like classic Orb as a result.  It's the last one.

I'm still thinking about ways to measure mastering ineptitude, so don't go away.  :-)

Measuring Dynamic Range

Reply #29 –
That's interesting. I don't know much about The Orb, but judging from commentary on Cydonia, could the change in mastering be more due to a more pop-like production style in general? I'm not sure if mastering alone can be fingered.

AMG also notes that Cydonia went through a tortured production process and that different producers were used on each track (with the exception of Patterson). One could hypothesize that because compression can squash out production differences between tracks, it was deemed necessary to make a consistent sound. So one potential experiment might be to try to find a correlation between the number of credited producers on a record, and the record's dynamic range.

Measuring Dynamic Range

Reply #30 –
Chromatix, have you considered adding an "album" mode to the Sparklemeter (so that an overall sparkle will be calculated for the album as a whole, in addition to the individual track-based sparkle values)?

It might appear on first glance that the overall sparkle would simply be the most extreme track value, although differences in individual track mastering could skew an expected result.

    - M.

Measuring Dynamic Range

Reply #31 –
Quote from: Axon on
So one potential experiment might be to try to find a correlation between the number of credited producers on a record, and the record's dynamic range.

Well, those results are going to vary wildly according to what each producer was actually given, and every one of them will have their own personal approach to mastering. Assuming each producer is handed the sequential analog results of all those who preceded him, then yes, the dynamics will be shot all to heck. But if a non-destructive approach is used, and each producer has the option of reverting to the original source or re-creating/utilizing part of a colleague's work, the whole presumption of cumulative degradation flies out the window! (For example, if they are utilizing previous work in a non-destructive digital system you'll still get cumulative colorization, but not necessarily any cumulative degradation of the overall dynamic, since the end-result is simply a direct application of the last cumulative edit).

    - M.

Measuring Dynamic Range

Reply #32 –
@Chromatix: Is there any news on the sparkle-meter front?

Measuring Dynamic Range

Reply #33 –
Quote from: Axon on
That's interesting. I don't know much about The Orb, but judging from commentary on Cydonia, could the change in mastering be more due to a more pop-like production style in general? I'm not sure if mastering alone can be fingered.

It's clear that Cydonia is not a good example of The Orb's work.  It's as if they're trying too hard - it isn't spontaneous enough.  It's music that's being pushed, not bounding over the hills of it's own accord.

Unfortunately, the next one - Okie Dokie It's The Orb On Kompakt - is even more pop-like, at least in terms of mastering.  The result is not pretty to listen to.

On the flip side, The Dream is considerably better.  I haven't yet analysed it by numbers, but it *sounds* that much better.  It sounds like classic Orb - quirky, fresh, well-layered, melodic, and with dynamic range you can actually hear.  The final track sounds like the kind of Orb that used to get into the Top 40 (an astounding achievement, given the kind of drivel that usually gets up there) in the '90s.


Quote from: weaker on
@Chromatix: Is there any news on the sparkle-meter front?

Not really - I've had to devote my attention to more mundane matters for the time being.  I haven't forgotten the project, but it is somewhat dormant for now.

I do rather suspect that I will have to rewrite it to use BS.1770 in appropriate places, and then re-tweak it to give intuitive results again.  I think this will successfully eliminate the bias on initial attacks from silence, and should in fact allow me to remove the artificial noise floor, without having any seriously detrimental effects.

Measuring Dynamic Range

Reply #34 –
Thanks for the info. Hopefully the mundane matters don't completely prevent you working further on that project :-D

Measuring Dynamic Range

Reply #35 –
Any chance of a quarterly "How is the project doing" update? This software was very interesting.

    - M.

Measuring Dynamic Range

Reply #36 –
It's also worth noting that the Pleasurize TT Dynamic Range Meter thingy is also coming out sometime soon. So now there are three dynamic range meters on our plate.

Measuring Dynamic Range

Reply #37 –
I've just looked up the Pleasurize site - looks good so far, if only because they're aiming at practical industry reform, not simply measuring.  However, I would hold off on commenting on their meter until they've told us exactly how it works.  So far I can't even find anywhere to download the beta, and then it would be a VST plugin, which I'm not entirely certain how I could use.

I haven't done any specific work on this recently, but I have thought of a good way to apply BS.1770 to it.  The really nice thing about 1770 is that you can "slide" the measuring window, once initially computed, by removing the leading sample and inserting the following one, which is much faster than re-summing the whole window.  The other huge advantage is that it is symmetric, since the ballistics are not causal - this lets it cope with silence (digital or otherwise) much more predictably.

I still think that there are three fundamental criteria to consider when determining the quality of a master: long-term dynamic range, short-term "sparkle", and ultra-short-term "corona".

I haven't come up with any sensible way to reduce the long-term DR to a single number, but it is relatively easy to interpret on a histogram.  I think research in this area should focus on producing a decent, standardised histogram.

The "sparkle" is most relevant on the loudest sections of the recording.  As such, perhaps a weighted average of the sparkle readings should be taken, with the weight depending on the averaged loudness in the vicinity of the reading.  This would neatly drop silence out of the equation, and focus the meter's attention on the areas where headrom is necessarily most limited.  Quieter sections don't matter for the sparkle measurement, because they are largely a matter for long-term dynamics measurement.

Likewise, the "corona" should be measured on top of the short-term sparkle samples, in the same way as the sparkle is measured on top of the averaged loudness.  The weighted average is also relevant here.

I'm thinking that the averaged samples should be taken over 1-10 seconds, the sparkle samples over 100ms, and the corona at either single-sample level or 1ms.  Any opinions?

Oh yes, before I forget, I even worked out how to handle file/sample edges.  The small sample window is normally centred on the larger one, but the sample windows never cross file edges, and are instead "sprung" against them.  So the first samples are taken hard against the file beginning, with the larger window remaining there until the smaller one has reached it's centre.

Measuring Dynamic Range

Reply #38 –
I haven't responded to this, primarily because I was confused. The  first time I read it, it almost sounded like you reinvented pfpf

Quote from: Chromatix on
I've just looked up the Pleasurize site - looks good so far, if only because they're aiming at practical industry reform, not simply measuring.  However, I would hold off on commenting on their meter until they've told us exactly how it works.  So far I can't even find anywhere to download the beta, and then it would be a VST plugin, which I'm not entirely certain how I could use.
It's up, referenced in the other thread.

Quote
I haven't done any specific work on this recently, but I have thought of a good way to apply BS.1770 to it.  The really nice thing about 1770 is that you can "slide" the measuring window, once initially computed, by removing the leading sample and inserting the following one, which is much faster than re-summing the whole window.  The other huge advantage is that it is symmetric, since the ballistics are not causal - this lets it cope with silence (digital or otherwise) much more predictably.
I'm not sure how much you're going to get out of point-by-point windowed computations. The performance advantage could go either way (having more stuff inside the loop might break memory coherence in some algorithms and make the whole thing MUCH slower). I don't think the symmetrical ballistics are that much of a good thing, because they're kind of psychoacoustically suspect.

Quote
I still think that there are three fundamental criteria to consider when determining the quality of a master: long-term dynamic range, short-term "sparkle", and ultra-short-term "corona".

I haven't come up with any sensible way to reduce the long-term DR to a single number, but it is relatively easy to interpret on a histogram.  I think research in this area should focus on producing a decent, standardised histogram.
Agreed. Like with all of these proposed numbers, there should be a method of rating the long-term dynamic range of a track that can be verified empirically to be most accurate through listening tests.

Quote
The "sparkle" is most relevant on the loudest sections of the recording.  As such, perhaps a weighted average of the sparkle readings should be taken, with the weight depending on the averaged loudness in the vicinity of the reading.  This would neatly drop silence out of the equation, and focus the meter's attention on the areas where headrom is necessarily most limited.  Quieter sections don't matter for the sparkle measurement, because they are largely a matter for long-term dynamics measurement.
I disagree strongly with this. This sort of mid-time stuff has a great amount of impact on quiet sections, arguably more than on loud sections. Think about a jazz track and how a drummer riding on a cymbal in a quiet spot, and compare that to a loud cymbal hit in a loud spot. I don't think it's safe to rate one sample over the other, if the added loudness happens to be the same in both circumstances. (There might be a psychoacoustic reason to do so based on ear nonlinearities, but that would almost certainly weight towards the quiet sections than the loud ones.)

It's risky to overengineer weightings beyond what is proven to occur in the ear. Sooner or later, if people wind up caring about these things, producers and mastering engineers will start working out how to "break" the algorithms, to give a track a far higher DR/sparkle/etc number than it deserves. Making more detailed rules and weightings increases the complexity of the algorithm, and may increase the potential magnitude of a flaw. I'd rather have an easy-to-understand meter that is a little wrong all the time, instead of a complex meter that is 99% correct 99% of the time, and is wildly wrong in that last 1%.

Quote
I'm thinking that the averaged samples should be taken over 1-10 seconds, the sparkle samples over 100ms, and the corona at either single-sample level or 1ms.  Any opinions?
The upper range is mostly dependent on the acclimatization of the ear. The sparkle needs to be resolving enough to understand the "beat" in music, ie able to handle a 120-180bpm song, while being long enough to respond accurately to bass frequencies (although those are always going to matter less after equal-loudness eq). The corona is always going to be more timbral in nature than dynamic unless it's longer than 5ms, but beyond that, there's a range between 1 and 220 samples it can take that really hasn't been defined yet.

At some point, we have to make a call. Are we aiming for a psychoacoustically accurate measurement of dynamic range, or are we developing numeric figures as estimators for mastering quality? The two are generally not the same, and focusing on very short-term responses is invariably going to work better for the latter than the former. But ultimately, I believe the former is more important in the long term, because it will better match listener experiences across a wider range of music, and will be less susceptible to producer rigging.

Quote
Oh yes, before I forget, I even worked out how to handle file/sample edges.  The small sample window is normally centred on the larger one, but the sample windows never cross file edges, and are instead "sprung" against them.  So the first samples are taken hard against the file beginning, with the larger window remaining there until the smaller one has reached it's centre.
That results in a noncausal response for the first few seconds, which I'm not particularly fond of. That said, my approach (preloading the first few seconds of music, iirc) is about the same thing.

So, how interested are you in helping develop a test for evaluating dynamic range meters?

Measuring Dynamic Range

Reply #39 –
I think I agree that different approaches are needed for an "anti bad mastering" and an "abstract" dynamic-range meter.  I will say that I've been concentrating on the "anti bad mastering" aspect, though I suspect that my original Sparklemeter does pretty well at the "abstract" measurement - but then so does a Dorrough meter.

Without researching the subject in exhaustive detail, I think that an "abstract" DR meter needs the following features:

- Localised measurement - that is, a graph over time, not a single figure of merit.
- Histograms for track or album summary.
- More than one pair of time constants.
- No distinction between loud and quiet passages.
- Psychoacoustically compensated response curve.
- Possibly take stereo and/or surround image into account somehow.

Whereas an "anti bad mastering" DR meter needs:

- Per-file and per-album single figures of merit, globally comparable.
- Only one or two pairs of time constants.
- Weighted towards loud passages; this is what mastering engineers are screwing up.
- No psychoacoustic compensation required.
- Consider channels independently or summed, imaging not very relevant.
- Robust against gaming.

I think it is possible for good examples of both kinds of meter to be based on the same fundamental algorithm (either simulated ballistics or BS.1770), but the post-processing will be very different.

I'll give you a concrete example of why weighting towards loud passages is a good idea.  Remember that one of my most dynamic tracks, as measured by Sparklemeter (v1), is from a Silver Classics recording of Berlioz.  Yet if I look at the loudest parts of that disc, I see and hear things that seem suspiciously like compression or limiting - a very consistent ceiling on the VU meter reading, a very harsh sound from some prominent instruments, and a general impression of "flatness".  Considering that recording technology was not at it's best at the time of performance, it is entirely possible that a compressor on fast settings was used to protect the equipment from loud passages, with the overall level set to reliably capture "silence".

Now, I do agree that a more complex specification is normally easier to find loopholes in.  But being too simplistic has the same danger.  With an unweighted measurement, the mastering engineer can exploit the method by using a moderate amount of long-term DR to increase the short-term DR, since on the quieter sections the compressors and limiters can be less aggressive.  Consider the potential "advantage" of starting the track 3dB down, and gradually increasing it to "lightbulb" level over the course of 3 minutes (which, approximately, is already often done).  Or he could gain a large unfair advantage by including lead-in and lead-out sections (which some musicians do like to include anyway), at a substantially lower level (and higher DR) than the rest of the track.  Or crowd noise - how many (modern) concert audiences do not clap, stamp, whistle or chant?

With the measurement weighted heavily towards the loudest sections, these problems are solved.  There is no benefit to the mastering engineer to reduce the DR of quiet sections - this doesn't give him more headroom where he needs it.  Instead it catches innocent mistakes (as I suspect in my Berlioz recording), and fairly detects today's "loudness war" efforts without being confused by anything that happens in quiet or silent sections.  Come to that, it should detect analogue "soft limiting" (such as overdriven tapes) as well as modern wizardry.

Measuring Dynamic Range

Reply #40 –
Quote from: Chromatix on
I'll give you a concrete example of why weighting towards loud passages is a good idea.  Remember that one of my most dynamic tracks, as measured by Sparklemeter (v1), is from a Silver Classics recording of Berlioz.  Yet if I look at the loudest parts of that disc, I see and hear things that seem suspiciously like compression or limiting - a very consistent ceiling on the VU meter reading, a very harsh sound from some prominent instruments, and a general impression of "flatness".  Considering that recording technology was not at it's best at the time of performance, it is entirely possible that a compressor on fast settings was used to protect the equipment from loud passages, with the overall level set to reliably capture "silence".
Even though such limitation was in place, isn't this track still among the more dynamic ones? How sure are you that the Sparklemeter was overestimating the dynamic range? Very dynamic tracks can still have large amounts of brickwall limiting. A few weeks ago we found that a lot of modern DG classical CDs were clipped, but they are still plenty dynamic.

Quote
Now, I do agree that a more complex specification is normally easier to find loopholes in.  But being too simplistic has the same danger.  With an unweighted measurement, the mastering engineer can exploit the method by using a moderate amount of long-term DR to increase the short-term DR, since on the quieter sections the compressors and limiters can be less aggressive.
I think most people would call that a quite tangible increase in dynamic range.

Quote
Consider the potential "advantage" of starting the track 3dB down, and gradually increasing it to "lightbulb" level over the course of 3 minutes (which, approximately, is already often done).

This sounds like a good thing too. In fact, that such a range even exists in the first place is more of evidence for artistic control over the process than a simple appeal to fad or commercial taste, and is harder to criticize.

In other words... if you're looking for failure modes of mastering - if you want to evaluate how long the music is redlining and how hard the compressors are being run/how distorted the music is/etc - I don't think you should even bother with weighting. Just throw out all the music that's below an extremely high loudness threshold, and have the results only apply to those loudest sections. Relatively few compression issues exist at levels significantly below the peak. That should tremendously increase your sensitivity to funny business at the peaks.

Quote
Or he could gain a large unfair advantage by including lead-in and lead-out sections (which some musicians do like to include anyway), at a substantially lower level (and higher DR) than the rest of the track.  Or crowd noise - how many (modern) concert audiences do not clap, stamp, whistle or chant?
This is a big issue, but when dropout filtering is employed (and pfpf does do this as does the TTMeter), the issue is minimized a bit, because the quiet sections that were never meant to be listened to at audible volumes do not influence the final result.

Most of the "loopholes" you're suggesting aren't loopholes at all; they make compromises in the loudness levels of the music that many producers would probably not be comfortable with. I'm not saying that big loopholes are not possible, but they are certainly easier to discuss compared to more complicated schemes. To some degree, we know more about them IMHO.

Quote
With the measurement weighted heavily towards the loudest sections, these problems are solved.  There is no benefit to the mastering engineer to reduce the DR of quiet sections - this doesn't give him more headroom where he needs it.  Instead it catches innocent mistakes (as I suspect in my Berlioz recording), and fairly detects today's "loudness war" efforts without being confused by anything that happens in quiet or silent sections.  Come to that, it should detect analogue "soft limiting" (such as overdriven tapes) as well as modern wizardry.
How would your weighting respond to multiplying the signal by a 20hz square wave, or more generally, pulse gating the music at high speed?

Measuring Dynamic Range

Reply #41 –
Let me put it another way - weighting towards high-volume sections is fundamentally the same thing as dropout filtering, just without an (exploitable) hard threshold.  Generally, hard thresholds are exploitable and inelegant.

Pulse gating is an interesting case.  Let's consider two scenarioes:  the small window being less than or greater than the pulse (or gap) width.

Clearly if the small window is wider than the pulses, then they will tend to be filtered out and will have no effect.  So at 20Hz and a 100ms window, that is what happens.

On the other hand if the small window is narrower than the pulses and gaps, then the samples will be recorded as alternately high and low DR.  The large window is unaffected because it is still much wider than the pulses, so these samples will have the correct weight.  So the weighted average DR will work out to be the same, assuming the averaging is done in log space (ie. in dB).

Therefore in neither case is the pulse gating a viable way of exploiting the weighted average.

WRT the Berlioz - I felt *something* was wrong the first time I listened to it.  But I don't go to real concerts often enough to be absolutely certain of what it *should* sound like - and the climaxes affected by the problem are the only places where directly comparable sounds are heard in that recording.  But everywhere else on the recording, it is excellent.  Having spotted the suspicious plateaux in Sparklemeter's graphs, I want to know more.

Measuring Dynamic Range

Reply #42 –
Quote from: Chromatix on
Let me put it another way - weighting towards high-volume sections is fundamentally the same thing as dropout filtering, just without an (exploitable) hard threshold.  Generally, hard thresholds are exploitable and inelegant.
Good point, I hadn't thought of it that way.

Quote
Pulse gating is an interesting case.  Let's consider two scenarioes:  the small window being less than or greater than the pulse (or gap) width. Clearly if the small window is wider than the pulses, then they will tend to be filtered out and will have no effect.  So at 20Hz and a 100ms window, that is what happens. On the other hand if the small window is narrower than the pulses and gaps, then the samples will be recorded as alternately high and low DR.  The large window is unaffected because it is still much wider than the pulses, so these samples will have the correct weight.  So the weighted average DR will work out to be the same, assuming the averaging is done in log space (ie. in dB). Therefore in neither case is the pulse gating a viable way of exploiting the weighted average.
So, you're right, but I'm still a little suspicious. One would need a much slower pulse gate to trigger any badness like this - like say 1hz - and at that point we are basically commenting on the ballistics of the long-term meter, which are tunable.

What I'm ultimately worried about (and maybe I'm changing the subject entirely here from weighting) is that low levels of pulsing, perhaps irregularly spaced, should show huge loudness increases at the peak levels if used in the right way, by modulating music far up in frequency into the most sensitive bands. I guess this would require gating periods in the 0.5-2ms range. This should show up with an equal-loudness filter, but probably wouldn't show up in anything that relies exclusively on ballistics or windowed RMS readings, and I think both the Sparklemeter and the TT Meter could be in that category. That is: if all you're doing is rectification, decay and/or RMS, you will not detect any of that funny business.

Quote
WRT the Berlioz - I felt *something* was wrong the first time I listened to it. But I don't go to real concerts often enough to be absolutely certain of what it *should* sound like - and the climaxes affected by the problem are the only places where directly comparable sounds are heard in that recording. But everywhere else on the recording, it is excellent. Having spotted the suspicious plateaux in Sparklemeter's graphs, I want to know more.
Again, going forward from here, I think the way to resolve this is going to be to attempt comparative testing of meter ratings of various samples against listener ratings of various qualities of those samples. We've got three dynamic range meters and it's entirely unclear to me which one is best. Granted, I'm biased.

---

Some more closing thoughts for tonight. Based on your descriptions of requirements for an anti-bad-mastering meter, I'm halfway wondering if some of the requirements (eg, clipping detection, brickwall limiting of peaks, distortion etc) should be separated entirely from the dynamic range aspect of the measurement. ie, the meter becomes sort of a design rule check for mastering, where grades are given to a sample for its various qualities, and a final composite score is given (either the average or minimum of the grades) which is the figure of merit. ie, it's clearly important to beat people upside the head for using brickwall clipping or digital overs, but it's going to be really hard to shoehorn that kind of stuff into a dynamic range meter.

On the other hand, regarding the "abstract" DR meter, my plan for pfpf for the last several months has been to move it towards giving histogram results across each critical band as opposed to across the BS.1770 windowed loudness. I strongly suspect that as we start delving deeper and deeper into this stuff, the very concept of loudness as a time-varying scalar value will become very tenuous - while limiting the measurements to wide critical bands makes design and interpretation considerably easier. Once the analysis is limited to a single critical band, a huge amount of psychoacoustic handwaving goes away. It also provides a springboard for implementing all kinds of wacky things like masking effects. That said, I have no idea how that sort of thing will turn out, and won't until I actually implement it.

Measuring Dynamic Range

Reply #43 –
Quote from: Axon on
We've got three dynamic range meters and it's entirely unclear to me which one is best. Granted, I'm biased.

I don't see a problem with 3 methods for calculating dynamic range. What would be nice is to have one program / plugin (foobar plugin would be nice) which could utilise all 3 methods, whereby users could either select one method/score or ask for a combination score based on 2 or all 3 of the methods.

C.
PC = TAK + LossyWAV  ::  Portable = Opus (130)
 

Measuring Dynamic Range

Reply #44 –
A pulse gating period of 1ms would be equivalent to 1kHz, which is slap bang in the middle of the most audible band.  I don't think they'd be that stupid, and it would still get averaged out very thoroughly by the 100ms window.  I don't even think you could argue anything useful about intermodulation effects - I know what amplitude, frequency and phase modulation actually do to a frequency band.

To be honest, I don't know enough about mastering to reliably put a "grade" on it.  I suspect, done properly, it's an art form like any other part of the creative process.  But I do think there are minimum standards that should be upheld, within which there is plenty of scope for artistry - which I suspect is what you mean.  Basic level, clipping and "climax DR" are relatively easy to measure, and it is plainly obvious when they've been done wrong.  Distortion is much harder, as it is often a legitimate part of the recording these days.