Error correction

May 4, 2022
 by Paul McGowan

I recently blogged about how building a better woofer in the first place resulted in a much better sound than adding a servo to correct the mistakes of a lesser woofer.

It’s common sense.

The same can be said for analog circuits where designing a low distortion, high speed step-response amplifier without need of feedback will almost always sound more open and lifelike than the opposite that requires massive feedback to fix its problems. (as a side note, feedback itself isn’t bad. In fact, it’s good if applied not to fix errors but to improve an already great performing amplifier circuit).

The point here is that in nearly everything one can imagine, it’s better not to make errors in the first place than to add a system that fixes them when they occur.

I’ll give you another example. This one from the recording studio.

Recordings depend on microphones to capture live sound. The better the microphone and its placement, the better the resulting sound. That seems rather obvious, right?

What I so often hear that makes me cringe is the idea of good enough “and we’ll fix it in post”. (post equates to the mix and editing process after the recording has been made)

Look at most big mixing boards (or their DAW electronic equivalent) and what you notice is an eye twisting bevy of knobs, buttons, and plugins designed to enhance that which has been captured.

I am sure you see where this is going.

It’s the purist’s thing.

If you get it right in the first place, there’s not only no need to fix it later but every reason on Earth not to.

The best enhancement in the world is to get it right in the beginning and then honor that perfect capture.

Error-free will always best error-correction.

Subscribe to Paul's Posts

45 comments on “Error correction”

  1. Makes sense!

    I understood that’s what the famous recording and mastering studios (Chesky, Grundman, Kevin Gray, Doug Sax etc.) did since the 80‘s to ensure their quality standard.

    I think those even did when they worked in the digital domain (where others tended to play too much with what was newly available).

  2. Indeed , it is always better to get things right in the first place than to fix it later. Another example opera singers and orchestras or solo piano players which are forced to adopt their performance (to fine tune their instrument/voice) to the given acoustic properties of the concert hall – no chance to fix things in the post. Unfortunately DAWs allowed to dramatically reduce production cost avoiding time-consuming fine-tuning in the first place. However today’s most advanced digital technology is able to fix “errors” in the post when it comes to phase and timing problem of loudspeakers and inherent problems (room modes) of listening room acoustics.

    1. Too often DAWs are used not to fix but to perfect and pull out humanity.

      Certainly Auto-Tune, pitching and pocketing (quantization) can be used to rescue sessions but they can also be used to turn humans into androids by enforcing perfection of performance no human can create.

      DAWs should be used for good, but too often they are used to homogenize and force music into a one size fits all template.

      Sadly, once we have a capability, the urge to use it becomes far too strong.

  3. I have to agree with this common sense post, Paul.
    I would think that those professional sound engineers
    who passionately care about their recordings would be
    the ones who strive for error-free & would actively shun
    I have very little knowledge about how a near-perfect
    recording is made; but I sure as hell know when I hear one 😉

    …& May the 4th be with you…sorry, I just couldn’t resist 🙁

    1. May the 4th indeed 🙂 It’s like Christmas, it comes round every year, but I hadn’t really twigged it this year, I was slightly distracted, which made the joke funny again, almost 😉
      But it reminded me of an important point that mistakes, errors, requiring correction, frequently happen when we’re distracted or not in the moment. I find that’s usually when I’m rushing, so it’s something I try to avoid. Chillax was the word, but it seems to have had its day.

  4. So you have an error free recording then as you suggested a few day ago. “I see no harm in adding a bit of spice here and there if it makes it sound better”.
    Isn’t this just a wonderful world.

    1. Indeed, Paul’s earlier comment represents the one decision every audiophile has had to make for themselves:

      If you have to make a choice, do you want to go for accuracy or what sounds good?

      For example, many systems, like Boulder’s, go for accuracy but are then described by many as cold and sterile.

      Other systems are warm and musical, but as with Franco Serblin’s original Sonus Faber designs, can be accused of making everything sound beautiful even when it should not.

      As soon as someone seasons to taste you are listening to their taste, but that has always been the case or there would be no need to seek out remasters by Steve Hoffman or Kevin Gray, or solid state or tube amps and preamps, or even our choices of cables.

  5. I do Front of House sound and though EQ may be needed on the output of the board in general the goal is to do as little as possible to the signal coming INTO the board.

    The nice thing is when you have the capability to fix issues when you need to, from a guitar pickup that needs a touch of warmth to adding a bit of highs to counter mic rolloff or cutting lows to counter proximity effect.

    But it never sounds as good as when the artists know exactly how close to be to a mic that with flatter response than a standard SM58.

    In short, GIGO is true of all audio situations, and mixer controls can be either the Febreze spray or the drenching of ketchup on the filet mignon of the audio world. 😀

  6. What if you don’t have the technology to get it right but you do have the technology in an invented servo drive to improve the sound? You add the servo device until you’re able to devise a better woofer that doesn’t need the servo.

    Another example would be you have bad knees so you put knee braces on them until a Doctor figures out how to repair those knees. You don’t stop wearing the knee braces because there is no treatment to repair them. Once there is a way to repair the knees you can ditch the braces.

    1. “You add the servo device until you’re able to
      devise a better woofer that doesn’t need the servo.”
      That’s exactly what Paul did; with Chris’ help.

        1. That’s primarily what I’m talkin’ about.
          Paul was forced to replace the drivers in the IRSVs because they were starting to disintegrate & Chris’ reasoning convinced Paul to leave out
          the servo.
          I’m guessing that if servo’s were still the best option that they would
          be in the FR30’s.
          Let’s see if servo’s make a comeback in the PS Audio subwoofer(s)

  7. Having the band together and all playing at the same time and you get the perfect take. That is rare, but when it happens it is great. On the other hand there is manufactured music. Sounds terrible, right? Maybe not.

    Even though I am not into jazz, I am a huge fan of Steely Dan’s jazz infused rock. Most fans of SD regard Aja as their best. I have about 4000 albums and I play Aja at least once a month. By the time “they” did Aja SD was just Fagen and Becker. They wrote the music and played their parts. Then they chose what they thought was the right session musician for each part that Fagen or Becker did not play. They recorded each individual session player and then created the final mix. In my opinion, they nailed it. It sounds real and engaging, the music draws you in.

    I think most bands do not go to this extreme. Most bands ( The Beatles, The Rolling Stones, Led Zeppelin, Bruce Springsteen and the E Street Band, Tom Petty and the Heart Breakers and others ) record as a band and then fix it or tweak it in the mix. Sometimes they do overdubs, but mostly they are a band.

    1. This is true, but what they did not do is go note by note through several takes and cut and paste only the perfect ones and combine them with perfect timing to make a frankentake if you will.

      However I think the original point still holds, the best end product starts with the best ingredients if you will.

      You can salt your taste but the results won’t be as good if you are adding flavors to counteract poor ingredients, and Steely Dan was always impeccable about the quality of those “ingredients.”

      1. According to an interview in the former version of Rolling Stone with Joe Walsh, Don Henley and Glenn Frey went note by note using Pro Tools on the Long Road Out of Eden album to make note perfect. 😮

    2. Ummm, just an aside. Jimmy Page used many, many overdubs in the creation of most LZ pieces. Songs like “Achilles Last Stand” have nearly a guitar orchestra on them!

  8. Paul, regarding feedback; “it’s good if applied not to fix errors but to improve an already great performing amplifier circuit”.

    Can you explain what is then being “fixed” if not errors?

    1. Sure. Maybe the wording is throwing us off. A quick example. Most IC op amps have such high gain that without feedback they won’t even work. And, even if they did, their bandwidth is miserable. Maybe 1kHz at best. Throw feedback around it and now the bandwidth increases dramatically and they work.

      As opposed to a circuit without feedback that works just fine but with feedback works a bit better.

      Feedback in a power amplifier improves rather a lot the damping factor. It would be an error to suggest it’s “fixing” a broken damping factor. A better word would be improving.

      Make sense? Instead of fixing an op amp that won’t even operate within the audio band without feedback, we are using feedback to enhance an already good circuit.

  9. I recall a late evening at Chicago CES in 1979. David Berning and Murray Zeligman were showing me the afects of negative feed back on the Berning Audionics BA150 amp. It had variable feedback from 15 dB to 0 Db. Each time the feedback was reduced the system sounded more like live sound. This bothered me since I knew reducing feedback raised harmonic and IM distortion and with the reduced damping factor it also created made the frequency response less flat.

    I knew by now that standard measurements didn’t always tell us how an amp performed. But here the measurements were worse and the sound was better. Finally I asked them if any measurement they did correlated to the sound expecting them to say none. But surprise both said the same thing, linearity. If one volt in produced 5 volts out the amp that was linear(say 2 volts in gave 10 volts out in this example) sounded the best, sounded more like the real thing.

  10. But of course, there’s no perfect microphone (or room, or instrument, etc) so there’s no perfect capture. Each microphone is different. You might use a tube U47 on a vocal, but not for drum overheads. So the tools, including the microphone, preamp, analog console, or DAW, and even the room, help enhance the imperfect recording and make it more perfect. Or in the case of some musical genres, mangle it into something it never was. Imagine “In the Air” without the extreme compression and delay. It’d be boring.

    1. Indeed. Depending on style of music, and what a musician is trying to convey, there are many ways to create a powerful and moving recording of music. If one is trying to capture the live sound of an acoustic band playing together with no sound reinforcement, that is one thing. But contemporary music can be much more than that, and the use of the studio as tool for helping to create a moving musical result is valid as well, electronic music being the most extreme example of such.
      Even in the case of just simple vocal recording, if it is being recorded in a studio environment, vocalists are not going to end up sounding “good” with out at least a little bit of compression and reverb of some sort. One might be able to record the same vocalist in a special space, taking advantage of the natural sound reinforcement (church, small well designed theater, etc) with careful microphone placement capturing natural reverb/ambience-but in the studio, a completely flat recording of a vocalist with a single mic is not going to sound very good without some “help”.

  11. Acoustic treatments in listening rooms is a great example of this. Instead of building as near a perfect room as possible, with proper dimensions and acoustic treatments, many will skimp on those and use room correction software to “fix the errors”. It will never work as well as a properly designed room because, among other things, room resonances are a time domain problem that room correction EQ can’t fix.

    1. I recommend to have a look at Trinnov Audio, France, and the company’s most sophisticated approach using a proprietary mic-array approach allowing to measure direct sound and “room”-sound! Very clever!

      1. Trinnov is excellent. I use Sonarworks, which is also excellent, and much less expensive. Instead of using a mic array, it uses a single mic, which you move around the room as directed by the software. The software determines the mic location by emitting a series of clicks from each speaker and triangulating the position based on the delay time of the received signal. Very clever! When you are done, it gives you a map of the room, with the various frequency responses in each location. Then you can average them any way you want and set the FIR filters based on that. It’s great for separating room modes from speaker frequency response variations.

  12. Your quip this morning reminds me of one of the first bits of audio wisdom I ever heard, which came from the founder of Linn: “garbage in, garbage out”. Ivor Tiefenbrun used this mantra to argue for the primacy of his LP12 as the first logical choice in assembling an audio system. I’m working on being less of a dualist, and can see that Ivor was right AND Paul McGowan is right concerning the importance of the loudspeaker in building a system. My 35-year-old Linn Sondek (updated, of course) sounds terrific through my PS Audio gear, which has yet to include the new PS Audio speaker. I have some terrific speakers, though, and know I couldn’t hear the Sondek’s potential without their resolving power.

  13. While it may be entirely correct that an error correction scheme via a feedback mechanism in a woofer is not as good as just making a better driver in the first place, I would suggest that this is not analogous to using feedback in electronics. The servo woofer example is many orders of magnitude slower in operation than feedback around a given analog electronic circuit. In terms of human perpetual abilities, feedback around an electronic circuit can be considered instantaneous.
    While I have enjoyed the sound from some audio amplification circuits with no feedback, I have also highly enjoyed the most recent work of Bruno Putzeys in class D (Purifi). The Purifi class D circuit uses more feedback that just about anything. Mr. Putzeys in discussing feedback has suggested that there is a bell curve in the application of feedback when it comes to sound quality, that is: just a little bit of feedback can sound good, a medium amount of feedback starts sounding bad, but as you add even more feedback up to very high levels the sound quality comes back again. To my ears the Purifi amplifier circuit solved the remaining “problem” I heard with previous good class D designs, the lack of natural, organic sound in the very high frequencies (cymbals sounding a bit more like white noise than a specific, realistic collection of discernible frequencies). In direct comparisons here, I have found no power amplifier in the four figure range which can compete with a good implementation of the Purifi class D design. I have heard some amps of class A/B that are their equals or maybe slightly better in some areas starting at around $16K. Given this listening experience, I tend to believe what Bruno Putzeys is suggesting regarding the application of feedback.
    Of course, these amplifiers first stage is critical to the sound as well-it might be interesting to see what happens, say if, the Stellar amps’ input stage (discrete, low/no? feedback) was connected to a Purifi output stage, instead of the fairly basic IC based first stage used in most Purifi amplifiers (even if “discrete” these input stages in commercial designs tend to be OPA designs with high feedback).

    1. I do not want to be super picky, but that class D performance curve as a function of feedback is a “U” shaped curve ( starts high, goes low and returns to high ) not a bell shaped curve. A bell shaped curve is the shape of a normal distribution.

  14. When its hot outside we use air conditioning. We do not move the house to another climate.

    When its freezing outside we turn on the heat. We do not move the house to another climate.

    When a performance is lacking something in how its heard in playback?

  15. I feel very sorry for Barry Manilow’s live sound engineers. He is clueless when it comes to mic technique and holds the mic over a foot away from his mouth sometimes. It’s horrible to watch and his sound is loaded with feedback.

  16. Servo tech is really something and whenever I hear the word it makes me think of my Cyrus CD i player. It uses a patented type of SERVO tech that allows for data reading of a CD gathered all at once to avoid using error correction. Cyrus feels that using error corrections in CD players creates time delays, improper transient responses and just too many sound anomalies.
    However I’m not naive. Some SERVO tech can be used for good and some for the bad.

    It is all in the hands of the engineer. 🙂

Leave a Reply

Stop by for a tour:
Mon-Fri, 8:30am-5pm MST

4865 Sterling Dr.
Boulder, CO 80301

Join the hi-fi family

Stop by for a tour:
4865 Sterling Dr.
Boulder, CO 80301

Join the hi-fi family

linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram