Could I trouble you to elaborate? I personally don't have knowledge of all the various parts and stages in AD and DA converters, but I have noticed, on playback, the quality (timbre, character..) of the audio often sounds less good to me if there are somewhat sustained sounds at -1 or -2 dbfs, even if there's no clipping. I always assumed it was the analog stage of the converters being pushed and not responding well. I have 3 different DACs here, and two of them perform noticeably less good at this loudness level than the more high end one I own. I've settled on avoiding crossing over -3dbfs, other than occasionally on very short transients, not necessarily because of fearing illegal data and clipping but simply because I assume especially a consumer playback device might have somewhat compromised performance in that upper range.

I can only describe the sound I'm referring to as 'pinched' and a bit strident ---basically, unpleasant. I wonder what might cause that, frequency imbalance? noise? filtering artifacts?

PS thank you for your generosity in explaining all this stuff.

Clipping an ADC was never a general practice. Mastering engineers found that certain very specific high-end converters could be clipped as a means of shaving a few peaks off. With most converters it didn't sound very good and, as Paul suggested, many converters sound best keeping peaks below -10.

Yep.

The obsession for printing to media at "absolute max" levels used to make a whole lot more sense with analog tape.

(Those old Ampex machines were brilliant in their ability to saturate the tape itself before any of the electronics started to saturate!)

...But these days (running at 24bit), you're really not going to go too far wrong with the following as a general guideline:

.

I've heard incredible sound peaking to -30!

What I hear and experience in the real world is sound.

I totally get what you're saying with regard to the data points that represent the sample values. And -- for sure -- the sample values without the crucial context of format are not signal by anyone's definition of the word.

I was more thinking of how Shannon uses signal in his Mathematical Theory of Communication. (Not that I am at all conversant in that truly seminal work, I rush to stipulate.) https://www.ece.uvic.ca/~agullive/shannon1948.pdf


Sorry for the distraction!

Hi Paul - sorry for the delayed response - I've been very unwell and off work since my previous post, but slowly coming good.

Actually, I find DSM indispensable for many tasks. I'm not even slightly disappointed with what it does. I'm disappointed that Plugin Alliance doesn't fix GUI and state-save problems in one of the company's most expensive plugins. Even if the problems are edge-cases, leaving them unfixed for years is an example of the 80/20 rule at it's worst.

Plugin Alliance have known about the curve problem for ages, but (like you) they don't know what causes it. Suffice to say, I cant shed any light on that - I'm not a developer. Maybe it's a permissions problem, but seems to me that should be easy to determine. People like me pay companies like Plugin Alliance to ensure the product does what it says on the tin. Beyond nagging them about it there's not much I can do except ask for it to be fixed.

As a mastering engineer, my biggest fear is reopening masters to revise them only to discover something in the chain isn't as it was when creating the master. In a few cases I've had revision requests come in many months (one was 18 months) after the project was originally signed off. I need to reopen projects in the knowledge the saved state reflects the exact settings from the previous render.

An aside: The difficulties of recall for analogue workflows is one reason why ITB mastering is now the default method for a large number of mastering businesses. Customers expect perfect total recall. Pricing is so competitive that ITB mastering is the only way some businesses can survive. Businesses which still use analogue workflows are usually long established, where the expensive outboard was paid for ages ago. It's almost impossible to enter the mastering business now with an analogue workflow unless the business owner has money to burn.

A friend who has the best analogue gear money can buy is really struggling with customers expectations regarding revisions and pricing. The market demands ITB and he cant compete by remaining OTB.

The good side of this lies with the available ITB processors. Nowadays, I reckon ITB masters are easily as good as analogue masters if the engineer knows what he/she is doing. I also reckon a pretty good argument can be made for ITB being sonically superior in many cases - processors like DSM simply don't exist in the analogue domain.
Ideally, move away from metallic 3D greyscale bitmaps to something more contemporary and consider scalable vector. Or, have a look at some more recent GUIs to derive ideas.

Like this:

TDR Nova | Tokyo Dawn Records

or these:

https://www.fabfilter.com/products/

The point being: In the 90s, Bob Lentini's SAW software sounded fantastic and was so stable it became the first ITB mixer to be used at live gigs. Bob rightfully earned industry kudos for building his app in Assembler and deriving industry leading sound quality and stability. But SAW looked terrible. Bob strongly resisted customer requests for a better GUI. This really hurt sales, to the point where Bob's stubbornness about his GUIs and things like VST support became the reason many didn't buy his products and his existing customer base fatally eroded. The market wants what the market wants. Customers are fickle because they can be. In today's audio software scene, there are so many good products that GUIs are a huge factor in customer choices.

I understood this before mentioning it and probably should have clarified. That said, it's not intuitive to have the curve below 1kHz compressed as is and I can't see how it could hurt the process to 'extend' the display. My brain wants to see what it wants to see. I don't want the processing algorithm to change - what I want is more complete visual insight/feedback from the UI. If I'm tweaking a low frequency via one of the parametric knobs, I feel a need for that to be displayed effectively in the threshold curve, even if below 1kHz is a single band all the way down to 20Hz. I have enough knowledge of hearing to know I cant trust my ears alone when mastering - good GUIs save time and reduce mistakes when tired. Quite literally, the best GUIs help us stay profitable and consistently professional.

If the threshold is being exceeded below (say) 800Hz, I prefer to know the frequency band/s provoking the threshold. It might be something I can deal with via another method before the audio reaches DSM.

Displaying a spectrum analyser within a plugin is trivial nowadays. I'm not sure if you spend much time checking out the latest GUIs in today's plugins, but suffice to say there are many beautiful options available.

The fact you've had to "explain it many times already" gives you some clues the threshold curve display is not intuitive for end users. Doesn't mean it's broken or really badly designed, but does mean it could be improved in line with what customers want. If that was sorted, you'd no longer have users asking why the threshold curve is so unusual compared to (say) typical multiband compressors.

I'm also perplexed, but it isn't only me and (AFAIK) it isn't only a problem within Sequoia and Samplitude. I could be wrong about that. Aside from problems years ago with Slate Digital plugins, I have zero problems with other plugins in Sequoia. I dumped the Slate plugins because the problems I had remained acknowledged but unfixed for years. I don't want to dump DSM.

NOTE: I have managed to workaround the knob problem by using the last version of DDMF Metaplugin, which is essentially a plugin chainer. The threshold curve isnt fixed within DDMF.

https://ddmf.eu/product/

JUCE seems to be a common problem with edge cases. When pressed, some developers admit the wait time for the JUCE developer to fix edge case issues is years, not weeks or months. I don't know if Plugin Alliance relies on the JUCE framework, but it wouldn't surprise me.

IMPORTANT NOTE: DSM works correctly in Reaper on my PCs, so it does seem clear that the host DAW software is a big factor. I'll raise it with Magix.

I'm going on the explanation from Plugin Alliance support, which indicated I'm not alone. I dont know how many users are affected. I also dont know how many users demo'd the product but abandoned it because of the curve bug. Plugin Alliance probably knows the demo vs sales statistics. Maybe you have demo vs sales statistics for your company before you handed the plugin to Plugin Alliance?

No need to be deeply embarrassed! I didn't say it is fatally broken (if it was, I wouldn't use it) and nor do I claim that sales are meagre, but my anecdotal observation is DSM rarely gets a mention when people discuss their tools. From that I extrapolated that the plugin could achieve better sales to up & coming engineers by revamping the GUI in line with recent competing products. That is all. Maybe it's a secret weapon!

Instead of being embarrassed, be glad that I'm here wanting to make DSM work better for me instead of using something else. No other developer has come up with a better alternative for some of the tasks suited to DSM - that is a reason to be very proud of your work.

Thousands of engineers? OK, didnt realise you'd sold so many licenses. That's great on one hand but perplexing on another, because thousands of $400 license fees should be a reasonable sum for a pro audio plugin developer in today's market. That said, no developer can survive with only one plugin as the company's only product, so it's not all that surprising you had to go down the Plugin Alliance path.

If I recall correctly, the plugin was originally $400 when purchased direct. The price became $329 when Plugin Alliance got involved. If you sold 1000 licenses prior to Plugin Alliance, that is $400,000, which seems pretty good for a single plugin.

I dont use it on every master because of the workflow issues. If those were sorted, I'd use it much more often. It is uniquely powerful and a great example of using code to innovate.

With regard to expanding into new markets, there's a real need for DSM in the iOS universe, which nowadays supports the iOS version of AU. iOS isnt only used by kiddies making beats.

If you retained any of the rights to the plugin, consider partnering with an iOS developer such as:

Kymatica | Software / AUM

Or maybe partner with DDMF - that company's products are available in iOS:

https://itunes.apple.com/us/develope...ag/id538434682

Needless to say, by the time we perceive the air vibrations as sound, it is no longer sound. Before we perceive it, sound is converted into something our brain can process.

Although I guess someone could argue that until we perceive sound, it's just vibrating air.

This is so true. The sound we perceive is generated in our heads from the stimuli we receive from our ears - and in fact even at the limit from our eyes and other stimuli too. It's an inner 'experience'.

This fact is not lost on the way we engineer processing to stimulate a desired effect.

Sound engineering is primarily an artistic process - and what we require from our audio tech is to provide us with the artistic tools we need and deliver the results in a reliable and consistent fashion. Audio tech should ideally be our slave - and not our master.

Much of the pseudo-tech from market creation and other commercial sources tends to undermine this IMVHO - because it entrains people in a kind of everlasting round-robin diversion from the real process of making art. In the middle of this confusion, people eventually seek out 'rules' to get what they need - which undermines the real need, which is to create art themselves, by whatever means at their disposal.

Some of the most successful art in modern music has been created by people who stuck their necks out and did what swathes of the 'intelligentsia' of the times would deeply frown upon :-)

If you ever want to have real fun, read Eric Kandel's "The Age of Insight: The Quest to Understand the Unconscious in Art, Mind, and Brain, from Vienna 1900 to the Present", especially the chapters on vision, and how the brain creates the image by searching quickly through its "filing cabinet" to create what it can't possible have time to "see".

It's the perfect book that melds art with science, written by a pulitzer prize winning neuroscientist and art lover.

https://www.amazon.com/Age-Insight-U.../dp/1400068711

I knew I should have said compression waves in free air pushing and pulling on my ear drums, sending waves through the fluid to...

That's a golden words, Paul.

Paul Frindle,
BTW, I thought I had a great idea for DSM but it didn't work out. I tried using two DSM's to process Mid and Sides (with help from Voxengo MSED) then recombine. It didn't sound good. The single stereo DSM sounded a 1000% better. Only tried it on one track. Perhaps that technology just isn't suited for M/S or perhaps it would require some internal phase compensation controller.

A quick addon to my previous post re: the uniqueness of DSM and my comments about the GUI - probably the closest plugin to DSM is Oek Sound's Soothe, followed by Melda Productions' Spectral Dynamics. In both cases the overall GUI isnt particularly exciting, but the curve display is more useful and informative than the DSM approach. The Melda plugin is scalable as well.

GUI design has come a long way since DSM was first released. Improving it could improve sales. Everyone here wants you to prosper Paul!

I also just want to add that I'm aware of course that this thread isnt an ideal place to discuss Paul's innovative plugin, but I came in at the tail end of the DSM-related diversion and wanted to contribute. Apologies for veering off topic!

This is an area where logic is often overruled by mistaken intuition and what people call 'common sense', particularly in analogue vs digital discussions.

The methods used by our body to capture and perceive sound are more like digital audio than analogue audio, particularly at the point of capture. Hearing is like eyesight - our body uses every possible shortcut to ensure maximum efficiency within the voracious electrical demands of the brain and nervous system. Our auditory and visual systems rely heavily on prediction and interpolation. We've evolved to see and hear what we need to survive, but the goal is speed, not accuracy. Hence, shortcuts.

If someone was put in charge of our auditory system design, he/she wouldn't install a tape machine into our skulls.......

...wowzers ...

thank you very much for your candour Paul.


Im a musician, a biology grad and I now work as an electrician studying electrical science to improve my trade grading

and I find endless fascinations in studying audio engineering and psychoacoustics (on my downtime) while studying electrical engineering ...

Knowledge of one feeds the other so much its like a series of eureka moments .

Electromagnetics vs, Binary code as a widely-held argument over which is better seems silly to me and so the analogue vs. digital (and consequent analogue emulation debate ) has seemed increasingly purposeless to me (no offence meant to those with strong opposing views on this)

IMO We should be forward facing like those electrical engineers in the last century who invented the ingenious things we still want to emulate now.

so for me, reading someone's opinion who was involved in the engineering of classic legacy audio engineering ...its priceless


Im only at the beginning of this thread , (page 4.)....

in the coming days....when I've read through and properly digested all the knowledge and experience in this thread...ill return with a question of my own if i may and i hope i get a complimentary "im now an audio genius " mug, t-shirt and keyring when i arrive back at this point .

...again...thanks for the candour and of course...respect to you for your great work in this field

many thanks

Dear Paul, I hope this isn’t OT but I purchased DSM when it first appeared, before Plugin Alliance existed. It’s truly an exceptional tool but AFAIK, there is presently no 96k version. Only 48k. Is this correct? Is there any possibility of a 96k version being released? Please excuse the interruption and take care.

As far as i can see, the DSM V3 supports up to 192kHz.


Cheers
Fred

Thank you Paul, i was misinformed. Take care.

This thread is incredible. Thank you @ paul

This thread is incredible. Thank you @ paul

I just fell into the rabbit-hole of rereading some of this.

I love this thread.

I would nominate the (I believe closely associated) study of our brains “Default Mode Network” as another worthy rabbit hole.



-Casey

Indeed! The one scary part is I don't think I'm as good at discussing this stuff today as I was a decade ago.

I learned quite a bit from this thread and Paul of course.

And I use DSM v3 in every mix.

As does everyone, I have more than enough diverse compressors, but oddly enough, I don't have an SSL Mix Bus compressor, and never really wanted one. Go figure.

Bob - I feel the same way sometimes. I've become somewhat disconnected from the sharp end because I'm now retired. But of course the principles are the same.

Paul, Bob - thank you both for sharing your wisdom even now.