REGISTER NEW USERLOST PASSWORD? WELCOME, Logout

Voicing

Paul's Posts — 21 July 2012

By

Voicing

Yesterday I happened to mention how we use certain elements to “voice” a circuit and that sparked a few questions.  What does voicing an amplifier mean?

Actually it’s a really good question because one would think you’d want to make everything as neutral and true to the music as humanely possible.  And, in fact, you would – but then reality gets in the way.

Let’s make a couple of broad generalizations first: most digital sources are relatively thin and bright while most turntable sources are slow and fat.  I know these are gross generalizations but when it comes to music, I don’t really have any other terms – and when you’re starting to design a digital circuit relative to a phono stage, you have to think differently about each.  Perhaps for this post accept the terms as at least relevant to the discussion.

Also accept that different devices have different sonic characteristics as do different circuit topologies.  Tubes and FET’s are generally warmer, softer and slightly big sounding, while most bipolar devices and topologies based on them are somewhat the opposite.

So imagine all these elements as having different flavors and different textures and you as a master chef.  You want to cook a world class meal and that involves combining all the various tastes and elements together to compliment each other and produce something remarkable.

This is what voicing is all about.  It’s probably a mistake when designing a circuit to pair a thin and bright sounding CD player front end with a similar sounding bipolar backend circuit – you’d be better off pairing it up with a warmer and softer sounding FET circuit, for example.

I know this isn’t very scientific but then, neither is music and the art of reproducing it in a way that reaches down to your core and resonates with your soul.

That takes a master chef.

 

email Voicing Forward to a friend and help us engage more readers

Get new and fresh stories like this each morning by joining the folks reading Paul's Posts. Click here

Related Articles

Rolling it off What’s all the fuss about? Young turks Get that mole!

Share

About Author

Paul McGowan

Paul McGowan is the CEO and co-founder of PS Audio Inc. a Boulder Colorado design and manufacturing company of high-end audio products and services. McGowan has been designing and building high-end products for nearly 40 years. Hobbies include skiing, music, hiking, artisan bread baking, kick boxing and cooking. He lives in Boulder Colorado with his wife Terri and his 4 sons.

(10) Readers Comments

  1. hahax

    July 21, 2012 at 5:59 am

    And then it also depends on the venue and seat in the venue what is the sound you are looking for. I recall Gordon Holt’s review of the KLH 12 speaker and the disagreement with Henry Kloss over the voicing. The next time Gordon went to the Academy of Music in Philadelphia he sat in a different seat and darned if that seat didn’t have voicing like the KLH 12. Is there only one ideal voicing?

  2. Gannon

    July 21, 2012 at 7:14 am

    This is another interesting topic…after having the benefit of meeting so many of the designers/engineers of some of the most amazing gear ‘back when’, I realized that with EVERY new product I had to inquire their procedure for finalizing it for market. Including the exact equipment used during their analysis. I got to the point where I could guess pretty closely how the product would perform using various associated gear, it was a fun game at the time.

    I could also recognize those with a larger, or more open and probably useful, approach…where they used a wide range of associated equipment to allow their designs to shine through MANY more real-world systems. I considered them more ‘mature’ in their understanding.

    There was also a huge divide between those who took a standard engineering approach, and those who allowed for some ‘mystical’ or magical “other-ness” that they could not define nor measure (yet)…but obviously made a difference. It is here that language fails…

    But it IS funny how many of those engineers would get all puffy and proud…when challenged by honest inquiry if there was anything they missed…something about the schooling or programming process filters away that portion of their psyches open to discovery.

    I am always honored when I meet those with engineering know-how who somehow maintain the innocent awe and wonder they had as children. Gives me hope, LOL! (geez, come to think of it, this happens in every so-called discipline…)

    Yeah, this is a curious topic…we need to get better at defining our expectations, and working towards agreeing upon language which can describe ‘em well enough for others to grasp.

    Cheers,
    John

  3. Soundminded

    July 21, 2012 at 7:29 am

    The timbre or tonality that characterizes a musical instrument or human voice was finally able to be understood thanks to the work of the mathematician Fourier. He showed us that we could look at the way in which sound pressure, electrical signals, or anything else where amplitude varied with time could also be viewed with equal accuracy as viewing amplitude varying with frequency and how to do it. Not only that but it didn’t matter if what we are considering is a repeating (periodic) waveform such as a sustained musical note or non periodic waveform that has no repeating pattern. This powerful tool is one of the pillars of our understanding about many aspects of our physical world and the things we build, even in part about how our own bodies work.

    The characteristics that differentiate one instrument or voice from another is the difference in the relative amplitude of different frequency components. It is why the note A which has a fundimental freqency of 440 vibrations per second sounds different played on a clarinet than it does played by a trumpent. It also explains some kinds of distortions of all types when these sounds are processed and reproduced by electronic means. There are linear distortions where the realtive amplitudes of the different frequency components are not in the same relative proportions as they should be. Lacking in lower frequency components makes them sound subjectively thin. Too much low frequency components relative to higher frequency components and they sound tubby or boomy. Too few high frequencies and the characteristic transient attack is reduced and the sound is perceived as soft, muffled, indistinct. Too much and it is harsh or shrill. These subjective comparisons are our reactions to this type of distortion. Then there are non linear distortions which are of two types. In one, harmonic distortion, there appear frequencies in the output that are not present or only to very a small degree in the input that are whole number multiples of what is present at the input. Intermodulation distortion and noise are a catchall for everything else that’s present in the output that’s not in the input. It’s a kind of mathematical garbage can where we put those deviations that can’t be explained by the other two categories.

    This method which helps us understand so much has its limits. Two things it cannot help us understand by itself are spatial and temporal distortions in sound. These have to do with the directions sounds travel and arrive in and their time of arrival. This is important because most of what is heard live are echoes that arrive from many different directions and at different times. For example, in a typical concert hall, sound dies out in a way that harmonics at 8 khz dissapear at twice the speed they do at 1 khz. in other words, the relative strenghts of the harmonics decreases at different rates. This alters the tonality of the sound. The initially arriving sound is the same sound you’d hear if you were listening to a performance out of doors with no band shell, no amplifiers, no reflections. It is ‘thin” and sharp. Then you hear the same sound again and again. As time passes the echoes make it sound mellower, richer, fuller and your brain integrates these sounds to give you a different picture of its tone. So what is the right frequency response for a sound reproducing system to be “accurate?” The answer is that unless you reproduce all of these echoes, whatever the frequency response that “voices” this sound, it is going to be wrong.

    The only known type of sound recording system that captures these echoes because the microphones can be placed where you would be listening is the binaural system. But the binaural system fails because it cannot capture the spatial components of sound and now I’m going to reveal why. When sound lands on a microphone the sound field is composed of vector components. Whether the sound comes directly from the instruments or from reflections it has directions of arrival. When it is converted to an electrical signal by a microphone it becomes a scalar. It has only components of amplitude varying with time. When it’s converted back to sound by a loudspeaker or headphone it’s converted back to a vector field again. The vectors that come from a loudspeaker are all wrong, every single time. Analysis shows that even reproducing sound recorded from a live instrument in the same room, a live versus recorded experiment, the vectors from the instrument have little in common with those from the loudspeaker and the difference is very audible. In trying to reproduce the vectors of a reverberant field of a large room in a small room from a single pair of loudspeakers the vector distortions are nothing short of monumental. While the miniature speakers of headphones are also technically vectors, since the direction sound arrives at your ears from moves with your head, from the point of view of your ears it is the equivalent of a scalar. That’s why the binaural system doesn’t work and can never be made to work. And it’s why you can’t reproduce the voicing of musical instruments using the types of sound systems we have today. In short the vector components reaching the microphones are not the same as those reaching our ears at a live performance and the vectors produced by the playback system just add far more of this kind of distortion even when the spectral response measures flat. Can this be fixed? Yes but it takes entirely different kind of sound system to do it, a system that is the result of analyzing and addressing these aspects of sound. No such systems exist within the purview of our technology yet. These problems remain entirely unaddressed. the paradigm used to understand it is inadequate. All of the thinking is still inside the box where there is no way to connect the nine dots with four straight lines.

    • hahax

      July 21, 2012 at 9:20 am

      how about a simple recording system with a stereo mike as primary(perhaps some extra mikes for subtle ambience) and then play it back on a speaker that disperses in an inverse manner to the mike. That should come closer(I’m sure not perfectly) to what you’re discussing.

      I recall Bob Fulton did all his LPs on a single stereo mike designed to be placed on the floor. And they were marvelous recordings both spacially and balance wise(he had a sensational ear as shown by his recordings and speakers). Too bad he never got to record a full orchestra. His recordings were mainly amateurs like high school choruses. In spite of that his recordings were often used as demo discs.

      • Soundminded

        July 21, 2012 at 9:38 am

        The first problem arises from the fact that unless the microphones are located where you would be listening and are omnidirectional you won’t capture the sound you’d hear at the spot you’d be listening at. Usually the microphones are placed much closer to the source of sound than you’d ever be and are often directional (cardiod pickup pattern.) Therefore they do not “hear” what you would hear. Some of the large room’s reflections do get into the recording but nothing like the proportion that you’d hear. Second, as I’ve pointed out they are not vectors during electrical processing but scalars. The directional properties of the reflections are lost. Finally not only is it impossible to extract just the reverberant components from the rest of the recording, their directional aspects (which are critical even though you are not consciously aware of them) are also lost.

        Why does this matter? Because it affects the tonality you’d hear live. If you ever get to hear a musical group in the open air without the benefit of reflective structures like a band shell and no artificial amplification, say for example an outdoor concert while the orchestra is tuning up or a small group who were invited to play at a picnic, listen to not only how soft and feeble the sound is even fairly close up but how thin it’s tonal quality is compared to what you’d hear indoors. The problem with voicing is that the tonal balance is inextricably linked to the reverberation, they are different aspects of the same phenomenon, that of reflections. The room literally amplifies the sound not just in amplitude but in time making it appear more powerful and changing its tone, its voicing. This is not a steady state phenenomenon and cannot be duplicated by a steady state filter. This was the mistake Villchur made in the 1960s justifying the high frequency rolloff of his AR speakers and the same mistake BBC made justifying their identical concept in the 1980s. Instead what you really hear is a first arrival of sound rich in high frequencies followed by a large succession of reflections which get progressively relatively less rich in high frequencies as their overall amplitude decreases too. If the reflections come frequently enough so that they are integrated by your brain as all being part of the same sound and not as separate sounds (the Haas effect) then their perceived tone or voice is affected by a spectral balance for each note that changes in time. To recreate that tone, you have to recreate the reflections much as you’d hear them live.

        • hahax

          July 21, 2012 at 10:39 am

          You’re one of the few people who knew that Vilchur rolled off his speakers. He hid it by giving the fequency response of each driver individually and each one was flat. But he didn’t make it clear that the level of the driver measurements dropped as you moved from woofer ot mid-range to tweeter.

          • Soundminded

            July 21, 2012 at 2:24 pm

            If you look at the graphs superimposed you’ll see that’s so, just as you say. Here’s an original ad proving your point though. This one is for AR3a but there’s the same deception. You can see many others on this web site for various models. It’s like a repository of everything known about AR an other speakers from New England of that era.

            http://www.classicspeakerpages.net/library/acoustic_research/original_models_1954-1974/original_models_brochures/ar-3a_brochure/ar-3a_brochure_pg2.html

            There are several points of clarification however. One is that with the level controls set a their dot positions, that was not the flattest response obtainable. That was obtained with both level controls at maximum. The dot’s indicated AR’s suggestion for flat based on the way they felt recordings were deliberately equalized. Also in showrooms dealers deliberately sabotaged AR demonstrations because AR treated them so shabbily.

            I experimented with a friend’s AR3s and they could not produce adequate treble. However, at two live versus recorded demonstrations in the 1960s that Roy Allison conducted, they seemed very close, infact sitting on axis the AR3 was slightly brighter than the guitar next to it. I didn’t get the explanation until a few years ago from those familiar with how it was done. Simple, Allison turned up the treble control on his Dynaco PAS3X preamplifier. AR can be equalized fairly flat especially now that we have the advantage of graphic equalizers.

        • July 21, 2012 at 6:03 pm

          Mark, that’s a really good point about the placement of the microphones relative to where you sit. Spot on actually.

  4. hahax

    July 21, 2012 at 5:35 pm

    Now if flat were the main component of sounding live. It’s certainly not the only one and so long as it’s not way off, I don’t believe it’s the only one. I would put dynamic linearity in 1st place. It’s necessary to seeming live albeit certainly not sufficient.

Leave a Reply