How do software plugins really work?

Discussion in 'Tracking / Mixing / Editing' started by godchuanz, Jan 17, 2006.

  1. godchuanz

    godchuanz Active Member

    Hi,

    I've been reading these forums quite regularly and have learnt a great deal from you guys, and hope to continue to learn more from you guys.

    There are a whole lot of software plugins, apparently some good and some not so good. The question is, if these software plugins work exactly to its prescribed function, e.g. the compressor compresses, the EQ boosts/cuts some frequencies..., then why do they differ? Software plugins are not susceptible to any randomness or external conditions, and work exactly the same way anywhere, everytime. So, if a compresor really just compresses, why is it that some are supposedly better than others?

    In the world of hardware gear, yes, all the confusing electrical circuits and quality of components can affect the sound, which is why some sound better than others (and are generally more expensive). But surely, in a computer, where everything is digitized and exact, 2 functions with the exact same parameters will always yield the same result?

    Let's take for example... software compressors. By all rights, a compressor will simply compress the audio waveforms according to the user defined parameters. If you set the attack to 10ms, the effect of the compression should kick in at exactly 10ms after the threshold paramerter is reached/breached, no more no less. It will then tame the ensuing sound samples by the ratio threshold parameter for a fixed period determined by the release setting. All the dirty work done behind the scenes by the software compressor are but programmed mathematical functions. So why is it that some mathematical functions "sound better"? Obviously, no software compressor will ignore the user input paramters and decide that its own parameters produce better sounds. So, given the exact same parameters for 2 software compressors, they should (by right) give the same result.

    And if we extend this logic to limiters, EQ and what not, most software plugins should be no different. Okay, except maybe plugins designed to emulate a certain sound (a certain kind of distortion, V-amp etc).
    So... why are there expensive software plugins claiming to offer better quality?

    Sorry if this has been discussed before, but I have tried searching and so far nothing really answers this question puzzling me for some time.

    Looking forward to hear from you guys.
     
  2. axel

    axel Guest

    I simply don't know!! and to be honest i don't care at all why, the only point important for me is to find units that sound good and do the required job.

    the only thing i guess is that programmers have different aproaches and levels of what they consider quality to be, just like hardware, design architecture... components, which seem not applicable for software, but maybee just the different approach of how to program something can maybee make a difference?

    it must be... fact is that there are little to rather massive differences in the quality and sound of plugs... i always wondered myself and in the beginning i even thought ahhh... it's all the same, software is software... untill i heard the difference!!!!

    i am not an IT / Programer type at all, so these are all guesses, except the fact that not all plugs sounding the same, not at all!!

    and that to be honest is the only thing i care about, but hey even in hardware you have to listen to it... in the end of the day your ears are the only judge, always when it comes to musical equipment or instruments...

    just my thoughts about it.
     
  3. chrispick

    chrispick Guest

    The short answer as to why plugins differ is that they do sound different from one another.

    To consider, let's take the compressor example you started:

    Compressors don't always "just compress." Often, they'll impart additional aural qualities to the signal (e.g., subtle harmonics), coloring what you hear. So, not only are they used to control dynamic ranges, they're also used to affect the sound. That is, they can be palette tools. And different "coloring" compressors add different "colors."

    Conversely, some compressors strive to be transparent and not color the signal. Producer and engineers uses both, depending on their personal tastes and what they think best befits the music.

    Moreover, different compressor plug-ins use different algorithms to achieve their task. Some are constructed to inflict minimal stress on your CPU; they allow you to load many instances of the plug, but at the sacrifice of sonic subtleties. Others incorporate more complex algorithms, taxing your CPU greater, but also offering more extensive, precise sonic calculations.

    Lastly, some compressor plugins have more bells and whistles than others. Some many only offer fixed gain and threshold knobs. Others may include addition functions like side-chaining, adjustable knee settings, multiband compression or channel strip abilities (like a simple EQ).

    These variances apply to EQs as well. Some are transparent; some are colored. Some are built erring to CPU efficiency; others are built erring to aural fidelity. Some have minor flexibility; others have more.

    Help any?
     
  4. godchuanz

    godchuanz Active Member

    Why will anyone design a compressor to do that? Forgive my ignorance, but I always thought designers/engineers try hard to preserve the sound in a compressor!
    But this is indeed a possibility. It just baffles me as to why they will want to do that. I will imagine every mastering engineer wants a totally clean, transparent compressor.


    If I'm not wrong, audio information is represented by samples (typically 44,100 of them in a second) lined up to represent (or approximate) sound pressure levels. It seems the only way to effectively alter the level of the audio will be to use a straightforward mathematical function of dividing the value of each sample by a given ratio.
    Since this is fairly simple to program, I would expect all algorithms built around this to consume similar CPU resources. Of course, I could be wrong!

    I don't know if engineers had indeed purposefully included coloring in their software plugins (which can consume quite large amounts of CPU resources). I think it's quite unlikely, but not discounting the possibility.
     
  5. godchuanz

    godchuanz Active Member

    Computers are consistent machines. 10 programmers will make 10 unique programs, yes. But the end result will be the same. Say... I want attack=100ms, release=100ms, threshold=-14db, ratio=3:1, gain=2db, every software programmed to do this, however differently programmed, will always yield the same result!
    That's why I am so puzzled.

    Exactly! That happened to me and got me thinking. I am a programmer myself and I thought about the algorithm behind simple plugins, but I simply cannot understand the logic behind it.


    You're absolutely right :)
     
  6. Kev

    Kev Well-Known Member

    I think you need to go back to the original premise

    what is a compressor ?

    show me a great sounding compressor and now tell me why it sounds as it does.
    ... write a mathematical model to correctly express what that favourite compressor is actually doing.

    seems easy but it is the original identification of what the compressor does that is not so easy

    The LA2 is a tube based opto compressor that makes use of the T4B electroluminescence device with two LDR's.

    When making one of these you may purchase 200 LDR units and reject more than 75% of them.
    What are we looking for ?
    and do the Programmers know what we are doing and why ?
    and do the programmers understand the full parameters of the circuit and who it responds to real program (music) material.

    Attack, Release and side is only the beginning of understanding compressors.

    Added to that there is no real standard for calibrating Attack and Release.
    ... even gain reduction is not a standard ... RMS or PEAK or somewhere in between.

    All of this is parallel to why Speakers or Mics with identical frequency plots don't sound the same.
    You need a far deeper and complex mathematical model to even begin to emulate these things.

    The Spice models for most of the Thermionic Tubes are WRONG and only an indication that a circuit will function.


    :roll:


    sorry but the tech in me is coming out
    I'll shut up now
     
  7. chrispick

    chrispick Guest

    Mastering engineers do want clean, transparent compressors, by and large (but not exclusively). Mixing engineers like a broader palette. The former is concerned with maintaining the intregity of a pre-existing mix. The latter is concerned, often, with creating an emotionally-captivating recorded interpretation of a song.

    For example, if you're recording, say, a jazz vocalist and you want to maintain the highest degree of aural integrity, but still need to control dynamics slightly, you'd reach for a transparent, precise compressor.

    On the other hand, if you're recording, say, a rock drum set and you want the snare to crack with cool overtones and the transients of the room mics to stand out more, then you might reach for a colored compressor (or two or more).

    To consider: Every rock song you've heard was made using compressors that affected the sound.

    Different compressors impart different sounds -- just like different guitar amps impart different guitar sounds. You wouldn't argue that there's no use in guitar amps sounding differently, would you? The same aesthetic -- and it is an aesthetic -- applies to compressors (and other studio gear), be they software or hardware.

    It's all about how you want to alter or not alter the sound. That is, it's about creative decisions. It's about purposefully controlling or creating sonic variance.

    Remember, original compressors were analog and prone to the distortions and signal irregularities inherent to the gear. So, our "listening vernacular," to coin a phrase, is based on popular music created with machinery that naturally affected sounds. Every pop song you've ever heard on the radio was processed, at least in part, by a color-adding compressor -- much of them had tons of it. In some cases, like, say, some of the later Beatles albums, this affectation was exploited and pushed to extensive lengths. And people liked the way it sounded (which is what counts in the end).

    Software compressor manufacturers are making products for people used to hearing music made with gear that treats a signal imperfectly. That is, they're making apps for everyone who's listened to the radio over the last fifity years. So, often they create algorithms that approximate the sonic irregularities of existing or obsolete analog gear.

    And, again, the same applies for EQ.

    Think of it this way. Harmonics and overtones are a big part of the music palette. It's the rare person who enjoys only listening to a pure sine wave. All those asymmetrical peaks and valleys in a waveform are what make it unique and appealing. Therein lies some of the complexity.
     
  8. ghellquist

    ghellquist Member

    godchuanz,
    there are probably a lot of reasons why the plugins sound different. I will only list some and make some guesses.

    -- of course, some are simply full of bugs. Not all programmers are created equal. If you start writing plugins yourself, your first try will have its load of bugs, at least all my programs has. My personal record is a three line program that did everything wrong.

    -- the "straightforward" "naive" (in a positive way) algorithm is not quite as easy to define as you describe it. Make a try yourself at setting up the exact parameters for running a compressor with attack/decay/ratio settings and with or without a soft knee. Different people will indeed handle this differently getting a different sound. And if you ever get into EQ-s there are a number of fundamentally different classes of algorithms. Most include phase-shift as a real-world EQ (simple algorithm), but some do without that (quite complicated algorithm), some even do it going over fourier transforms (now we are talking major CPU load).

    -- even if you get the algorithms mathematically correct, you have to handle numerical error propagations. Some "correct" algorithms can become unstable unless you carefully take care of the way rounding will effect your result.

    -- the "straigthforward" algorithm might simply sound bad. After all, the ear is the decision maker. Once you have tried your plugin on a number of sources you might end up realizing that the result simply stinks. No life, pumping, boring are things that you might end up with. Here is the starting point for changing things using your ear, to make it sound right, not necessarily doing things the "theoretically right way".

    Regardless, go ahead and write some plugins, you will probably have great fun.

    Gunnar
     
  9. axel

    axel Guest

    Chrispick wrote:


    ohh, this is a very valid point...

    godchuanz wrote:

    to give you a choice and some character... thanks for that.

    my personel fave compressor plug for example the Blockfish by digitalfishphones... is everything but neutral it's full of colour... and that's what i love about it... i whish there would be more stuff with personel character / sound out... all this "neutral" stuff... horrible!!

    we would end up sounding all the same... nearly :D,

    that's why it is so important (and i don't want to miss this!!!) to choose from a wide varity what you want... and i need both... sometimes i want something neutral to have more control and sometimes i like lots of colour in the sound.... i simply want CHOICE!

    axel wrote:

    maybee you can e-mail sasha eversmaier from http://www.digitalfishphones.com, he is a nice guy and programmer at magix (samplitude / sequia) and responsible for bringing us those wonderfull NOT NEUTRAL SOUNDING PLUGS the fishfillets and ask him!!!!!

    cheers
    axel
     

Share This Page