Is it Live? Or is it . . .

As someone who started out as a classically trained musician but who then stepped sideways into rock, I'm fascinated by the one music I've never played: jazz. It seems to me that the essential difference between a performance of a classical work and a jazz performance is that in the former, the musicians use their technique to breathe life into dead notes on a page, while in good jazz, the performer not only applies a similar level of technical expertise, but also has simultaneously to have all of music theory at the fingertips in order to decide what the next note should be. It is a rare musician---Keith Jarrett, for example---who can excel in both arenas.

All of which is by way of introducing a topic I briefly discussed in this space last December: The fact that live sound differs from recorded not just in degree but in kind. This is especially true at high frequencies, where live highs may seem suppressed in level compared with recorded, but have more delicacy, more "shimmer."

Everyone has had the experience of unexpectedly hearing the sound of musical instruments and immediately realizing that they are real. The stimulus for this column came a few months back when I was attending a conference on Digital Signal Processing. I was walking through downtown San Jose with two friends and heard the sound of two electric guitars playing jazz. The fact that they were playing live was unmistakable, yet---and this was the kicker---the instruments were electric, all the "real" sounds being produced by loudspeakers!

So what on earth can be the readily identifiable difference between the sound of a loudspeaker producing the live sound of an electric guitar and that same loudspeaker reproducing the recorded sound of an electric guitar?

Part of the difference must surely be that, even with 20-bit digital recorders capable of capturing an electrical signal with a fidelity unheard-of even 10 years ago, the sounds on every recording still have to be captured with microphones---every one of which changes tonal colors to a greater or lesser extent, and introduces its own set of spatial distortions depending on how it is used.

Many point out that dynamics play a key role in defining the difference, that many engineers routinely apply compression when they make a recording. It's true that the subtle differences between pianissimo, piano, and mezzo-piano are generally clear in a live performance to an extent hardly ever experienced from a recording. But the lack of compression can't be the determining factor; many purist recordings are made with relatively uncolored microphones that appear to capture the true dynamics of live sound, yet still sound "recorded."

I was given a clue to this puzzle over a decade ago when, as part of a talk I was giving at a London Heathrow Hi-Fi Show, I played my Fender bass guitar through a pair of Linn Isobarik speakers. The Isobariks were widely and correctly regarded in those days as having excellent dynamic range. When they were fed the direct-injected signal from the Fender bass, however, the woofers---four KEF B139s---crapped out at a surprisingly quiet level; surprising, that is, compared with my experience of the speakers reproducing recorded bass guitar.

The reason for this unexpected accident was simple. Though the average level of live and recorded sound may be the same---and it is the average level that determines "loudness"---the "crest factor" of live sound is greater in that the peaks are much higher in level, much "spikier," than they are once recorded and played back.

What is it about sound reproduction that reduces the transient spikiness of musical waveforms? As the English audio engineers Ben Duncan and Laurie Fincham (the latter once with KEF but now the designer of the excellent high-sensitivity Infinity Composition Prelude loudspeakers) pointed out some time ago, all recorded music has passed through one, more, or many cascaded high-pass filters---filters that roll off the low frequencies. And the sound of a high-pass filter is unnatural in that, as far as I can ascertain, it doesn't exist in real life. All natural acoustic filters roll off high frequencies---they are low-pass filters. There is also the fact that the in-band phase shift from a high-pass filter is much more severe than with a low-pass filter in that it extends much further away in frequency from the filter's nominal cutoff frequency (try modeling them sometime with SPICE or some other circuit-analysis program).

The absence of high-pass filters with live sound means that the musical waveform is preserved. The presence of high-pass filters in all reproduced sound means that the waveform is never preserved. In the words of Kalman Rubinson, an occasional contributor to The Audio Amateur and a member of The Audiophile Network, (footnote 1), "Something in Nature abhors a capacitor."



Footnote 1: Tel: (818) 988-0452 (modem, 1200-28,800 baud, 8 bits, no parity, 1 stop bit), (818) 780-6260 (fax) or (818) 782-1676 (vox). Web: www.tanet.com .
Share | |
Site Map / Direct Links