As someone who makes a living creating CD masters for record labels, and an equipment designer also, I had to send a response to the article by Mr. Giesberts.
The timing of the arrival of the 1/99 issue of AE was interesting, since I was working on an album that contains tracks mastered both here and at perhaps the best-known mastering facility on the West coast. I was in the process of examining peak levels. This topic also relates to a page called TechNews that I'm about to open on my website, www.drtmastering.com.
For major and most independent record label releases, it is mastering facilities, not recording studios, that are responsible for setting the levels that you find on a CD. This is a sector of the industry that is not well known outside the business itself, and there are a some common practices worth mentioning. It is very common to have clipping on CDs, in all music styles, but especially in Pop. This problem takes two forms:
1)Actual clipping - the waveform at the maximum positive or negative value for two samples or more.
A little history: The Sony PCM 1610 and 1630 processors, writing a video signal to a 3/4" Umatic tape deck, produced the vast majority of master tapes for CD production through the '80s and early '90s. These had a clipping or "Over" indicator that would light up at one of three presettable levels: 4, 8 or 16 consecutive samples at the positive or negative rail. The default setting, which most professionals used, was 8 samples. This means that you could clip the signal for 7 samples and no "Over" would be indicated. Folks into more aggressively hot music could set the limit at 15 samples. Depending on the program type, and what compression was used just below that Over threshhold, Sony and others determined that clipping on this scale was not a problem for most consumers. So, like it or not, there has been clipping on CDs from the start. If it was always restricted to below 8 samples, or even 16, that would be the good news....
(I have to comment here on the editor's parenthetical remark. It is precisely because of commercial acumen that label execs and producers demand that their material be recorded as hot as possible onto CD. There is an unfortunate tendency in the industry to want your CD louder than the rest, because it gets people's attention, at least for a short while. Any mastering facility or studio that refuses to record a hot level for a client who demands it, and (trust me) many do, will lose that client.)
And now, the bad news....
2) Clipping below maximum possible level.
Mastering engineers who are asked to cut a hot level (and who want to keep their jobs) do the sensible thing: They insert a limiter ahead of the CD burner that will restrict the maximum level to perhaps just a sample or two below the rails. Since there are roughly 65,536 discrete levels that can be represented by a 16-bit number, throwing away two or four is not a big deal. Dynamic range is not affected noticeably, the CD replication plant will accept the master as having no Overs, and will cut the disc without further signal processing. (The fact that plants often do processing is yet another can of worms). Since the disc is technically within spec, now the clients are free to insist on even hotter levels for their next album. This means that more of the disc will be clipping, but now at a "safe" level just below the rails. It is not uncommon to see MANY dozens, even hundreds, of consecutive samples clipping in this way, depending on music style.
The only silver lining is that if you apply the proper type and amount of compression and limiting, so that the transition into (moderate) clipping is smooth, many people actually enjoy the sound of the artifacts that result. Harmonics are generated, the sound gets momentarily brighter in character, and people think they are hearing clarity, rather than distortion. (Consequently, getting this transition zone right - the upper few dB of dynamic range - is where I continue to spend a large amount of time when designing mastering gear. In Audio Electronics [magazine] tradition, its possible to build equipment that outperforms anything you can buy.)
This is big business for hardware manufacturers. They sell hard limiters, soft limiters, tube and solid state, compressors with hard and soft knees (transition zones). One manufacturer of A/D converters used on tens of thousands of CDs builds a "soft clipper" right into the converter, with the express purpose of making the music sound louder. You can also buy multiband digital processors or software for your PC which will anticipate when the signal is about to clip, and reduce the level before the clipping occurs. Radical er... waveshaping (don't call it distortion - a bad thing) may be required to do this, but the result is that you can get crushing average levels onto the CD.
Does this sound good? ummmmm...... welllll........ . its loud. What to do about all this? Demand more conservative levels? Don't buy CDs with clipping on them? Change the record labels' practices? Perhaps all of the above....
Monitoring the signal with an oscilloscope is very informative. If readers are interested, I can suggest a very simple circuit ($5.00 in Radio Shack parts) that you can insert ahead of the scope, doubling the vertical resolution and make it much easier to read. I use it as one of my visual monitors when working on CD masters.
Keep up the good work. David Torrey DRT Mastering 603-924-2277 info 603-924-6845 priority