How's That Sound?- Digital Filtering

How's That Sound?- Digital Filtering

Posted March 2011

Renowned record producer Mark Rubel of Pogo Studio, recently spoke with Legacy Audio President and chief engineer Bill Dudleston about A/D converters, CD's, and recording practices.

Here's what they had to say about changes in recording quality over the years.

Mark Rubel: I think it’s a couple of issues, which one could separate into technology and practice:

First, the early A/D converters just didn’t sound good. Bad-sounding circuits designed by computer engineers, not audio engineers, and especially the “brick-wall” filters designed to keep out information beyond the range of hearing. Some of the early machines used analog brick-wall filters, which got into the audio band and caused phasing problems. It’s also arguable that analog recordings such as tape and vinyl can capture frequencies beyond the upper limits of our hearing. Though we don’t hear those frequencies, we can hear their interactions and interferences. If for example 22khz and 23khz exist in the original source, though we wouldn’t hear those frequencies we could hear the sidebands and interference frequencies, which should include 1khz- the difference between the tones.

You’ll also notice that a lot of original recordings from the ‘80s, not just reissues, had a tinny and harsh sound, and to an extent that may have been the aesthetic at the time, which people may have emulated in mastering to try and “modernize”. Maybe it was the boom box, or the beginnings of the “loudness wars”.

As for the digital recordings of the time (as opposed to remastering of earlier recordings), I think it’s a combination of tools and technique. Those early digital tape machines, though very expensive (around $250,000 for a Mitsubishi 32-track digital machine) did not sound good. The filtering was bad, the analog circuitry was not good. Though, the rumor is that the early 3M digital decks sounded amazing (16-bit, 45ips, 50khz)- Donald Fagen’s Nightfly was done on it, and it sounds good. But still, it was $150,000

So, here’s an interesting point regarding technique: at that price point, and given that most recording was then done in professional recording studios, the people who were using this new recording technology would have been experienced professional recording engineers. That means that, like me, they would have only have worked on analog tape up to that time. Many of the sounds and techniques that had been developed to that point were based on analog recording, and whether consciously or not engineers would usually record things with extra treble to counteract the loss of high end that came from tape wear and self-erasure. Condensor microphones tend to have a high-frequency resonance that also complements the increasing dullness that comes from the total tape wear, head wear, calibration anomalies, self-erasure and decay of the tapes etc. This accounts to some extent for the much-vaunted “warmth” of analog tape, which could be a nice term for “dull”...

Even despite the previously mentioned audio shortcomings of early systems, at best a well-designed digital recording system is a mirror that doesn’t add a sound of its own. So it’s understandable that there was a significant transition period that fits pretty well with the era that you mention, from recording engineers using the time-honored techniques that they’d developed for tape recording with digital, realizing that if you send it an extra bright signal you’ll get it back, and then gradually adapting their techniques to the new medium. There’s a movement in recording towards ribbon microphones, which is actually an earlier technology especially advanced by RCA in earlier days. I think it’s the same thing: condensors worked for analog, and ribbons’ smooth or de-emphasized high end complements or counteracts digital recording in a pleasing way.

Also at the beginning of the ‘80s there was a move in electronics towards integrated circuits and surface-mount technology, and away from tubes and discrete transistor technology, where the components tend to be more separate and large, better built, better able to disperse heat and tolerate higher voltages. Partly this was a result of the corporate mergers and a movement away from smaller companies that specialized more in boutique and custom-built equipment, that tended to be run more by people who care and are discerning about audio quality. When the audio gear is made by companies like Sony, 3M and Mitsubishi, the decisions tend to be made less on the basis of fidelity, and more on the basis of cost, mass production, etc. That sort of equipment was more likely to be designed by computer engineers than audio engineers.

Bill Dudleston: Digital filters are best but need to be employed properly.

There were so many things going on when digital came on the scene. Audiophile company's like Denon made better players with quality analog stages, but there was a lack of digital masters to show it off. They even tried to make their own recordings. They admitted to aliasing filter anomalies. Jack Renner at Telarc made sure their releases emphasized dynamic range, but the strings sounded etched.

Tweeters were evolving and the textile domes where getting much better, but the metal domes all had a harsh ring between 16kHz and 22kHz. B&W introduced higher order crossovers that exhibited huge Gibbs energy at the corner frequencies. The lobing error was reduced but the complex impedance was severely contorted. Graphic EQs might overcompensate a mix with a high Q boost in the 3kHz range on two ways.

Monitor quality was all over the place. JBL, Rogers, Yamaha, KLH, B&W, ADS, Bozaks, Spendors all sounded so different. When my friend Steve Hoffman remastered Jethro Tull's Aqualung he called Ian Anderson to find out what monitors they used when mastering because the tonal balance was so far out. Ultimately, the NRC and the Canadians forced manufacturers to build more accurate monitors if they were going to compete against the API group of manufacturers.

The thing that bugged me the most about early digital was a large decrease in L-R component below 1100 Hz. This decrease in the out of phase component and the pitch black silence was the difference between painting on canvas and flat glass. It was sterile, lacking depth and ambience.

It is for certain- 24 bit 96Hz through a Sharc or Sabre is far more accuracy than needed to reproduce recorded music. We now have an essentially perfect recording medium. One thing that amazes me to this day- artists will play a recording they know intimately over a speaker such as our Helix and state that only now had they heard the recording for the first time. This suggests to me that loudspeaker variation is still a limitation in the process.

What do you think about the changes in recording quality?

Mark Rubel

Since 1980, Mark Rubel has made about 1,000 recordings at his Pogo Studio in the center of The Center of the Universe (Champaign IL), where he has recorded Hum, Alison Krauss, Rascal Flatts, Adrian Belew, Melanie, Luther Allison, Henry Butler and many others, including work for RCA, Capitol, Warner, Jive/Zomba, etc. Mark belongs to SPARS, AES, NARAS, ASCAP, the Music and Entertainment Industry Educator's Association (MEIEA), Chicago’s Engineering and Recording Society (EARS), and many other acronyms. He is a member of the Grammy® Producers Committee, the NARAS Producers and Engineers Wing and Education Committees.

Having taught audio at the college level to more than 4,000 students since 1985, Mr. Rubel now serves as Audio Director for Eastern Illinois University’s $65 million Doudna Fine Arts Center, where he is lucky enough to teach Audio and Recording Technology, the History of Rock, and works on developing new audio and music business curricula.

Mark travels about doing panels and workshops, including a 2010 AES panel with Al Schmitt, Ed Cherney and Elliot Scheiner. He has been featured in various books and magazines, interviewed Les Paul, Terry Manning and others for Tape Op Magazine, and acts as a consultant, beta tester and audio expert witness, mainly in the area of copyright.

Mark’s thrash oldies band Captain Rat and the Blind Rivets is now in their 31st year. He plays many instruments badly and sings worse, and is ridiculously happy cultivating songs, students and cats, along with his beloved and saintly wife Nancy.

Bill Dudleston

Bill Dudleston is President and founder of Legacy Audio, a high performance audio and home theater equipment manufacturer located in Springfield, Illinois. A member of the Audio Engineering Society and the American Institute of Chemical Engineers, Dudleston was recently inducted into a regional Hall of Fame, taking his place among mid-western Nobel scientists and statesmen. He is listed in Who’s Who of American Businessmen.

University of Illinois graduate and inventor/ patent holder of numerous circuit topologies and acoustic alignments, Dudleston has pioneered controlled directivity loudspeaker designs, wave-launch coherence in low frequency radiators, dynamic braking in active speaker design, selectable directivity multi-way microphone arrays, feedback eliminating stage monitors, and isolated wall-mounting methods for in-wall/on-wall speaker systems. His innovation in business practice, customer service and technology is noted in Tom Pettsinger’s The New Pioneers. Dudleston has published numerous articles on acoustics and loudspeaker design. He also authored Reinforcement, Resonance, and Reverberation: Fundamentals in Sound Control.

Bill Dudleston has designed and provided Legacy speaker monitors for Arista, Sony, Universal Music Group and archival organizations such as the Stradivari Violin Society. Multi-Grammy award-winning producers Rick Rubin, Antonio “L.A.” Reid, and renowned mastering engineer, Herb Powers, have utilized and publicly touted the Legacy designs as assisting in producing artists Sheryl Crowe, Johnny Cash, Tom Petty, The Red Hot Chili Peppers, Mariah Carey, and Usher. Re-mastering engineer Steve Hoffman has utilized the Legacy speakers on re-issues of Elvis Presley, Frank Sinatra and Nat King Cole.

Legacy Audio and its founder have been cited in publications such as Billboard, the Wall Street Journal, Stereophile, The Absolute Sound, Home Theater Magazine , The Robb Report, and The Sensible Sound. Dudleston continues as an innovator in the areas of DSP, digital amplification and wave-launch reconstruction while directing Legacy Audio’s research and development program.

Further information on the Gibbs phenomenon.

The Gibbs phenomenon is the peculiar manner in which the Fourier series of a continuously differentiable periodic function behaves at a jump discontinuity The overshoot does not die out as the frequency increases, but approaches a finite limit.^{^1}

All higher order filters, active or passive suffer from Gibbs energy. When realized digitally we have learned to use FIR filters (finite impulse response) that do not ring indefinitely or by employing a smoother method of Fourier series summation or by using sigma-approximation. Using a wavelet transform with Haar basis functions, the Gibbs phenomenon does not occur at all.

1. H S Carslaw (1930). Introduction to the theory of Fourier's series and integrals (Third Edition ed.). New York: Dover Publications Inc.. Chapter IX.

[1]

Remember at 44kHz sampling rates, there are only two samples to represent a 22kHz waveform.