Recording - July 2010 - 6

Note to faithful writers: When sending your letters by email, please remember to include your name and where you hail from.
Audio file quality—how much is “enough”? Which method produces the highest quality recording, to record at CD quality, 16 bit and a 44.1 sample rate or to record at a higher bit depth and sample rate and then convert to CD quality/MP3 format? I am not concerned about the additional memory it will take to store recordings made at higher sample rates, and I know that when recording at higher sample rates, it is best to record using rates that are multiples of 44.1 for better conversion.
Jerome Tarver via email

Jerome, there are lots of opinions on this subject, and very few hard and fast rules. But let me give you some pointers on areas you’ll want to consider and test for yourself, letting your ears be the judge. In my opinion the resolution (bit depth) of the music you record is a no-brainer; you want to convert analog to digital at 24 bits. In the early days of 24-bit converters, a lot of what you actually captured was noise (we called them the “marketing bits”), but a good set of converters now can give you about 22 bits’ worth of usable dynamic range. If you go by the rule that each bit of added resolution gives you 6 more dB of usable dynamic range between the loudest and softest sounds you record, then 16 bits gives you 96 dB, which seems like a lot until you realize that most rock concerts are way louder than that and really loud sources like kick drums can be well over 120 dB SPL. 96 dB is actually a very usable dynamic range if you maximize your use of it; you want your loudest sounds to be right up against 0 dB full scale, so that your softest sounds aren’t down at levels where they’re only being represented by two or three bits of data. This is the real reason to push loudness to the max on CDs... but note that we’re talking about peak loudness here, not average loudness! If your peaks are near 0 dBFS, you can still have a lot of dynamic range... but you run the risk of clipping if you have a particularly loud peak. There’s not a lot of headroom, and little margin for error. When you’re recording at 24 bits, those extra 8 bits extend your dynamic range to 144 dB; in practical terms that means that if your loudest recordable sound is the sound of a guided missile launch at close range, your last couple of bits will still be recording electronic noise floor. You’re up against the laws of physics at that point. You don’t have to hammer your levels, and will still have lots of dynamic range without pumping up your noise floor—even if you set your levels conservatively so as not to clip anything. Note that this is different than the bit depth of audio inside your DAW. You want that to be as large as possible, because every time you perform a DSP operation like mixing or adding effects (and you’re doing that tens of thousands of times every second), you’re doing math, and rounding errors get shaved off... the more bits you work with, the more accurate your result. That’s the real benefit of a 64-bit data path inside a DAW. But conversion, analog to digital and back, can be done at 24 bits and will satisfy all but the most hardened fanatics.

Sampling rate is a larger controversy because many people claim that 44.1 kHz is perfectly good for what they do and that higher sampling rates don’t really gain you anything in terms of sound quality. In my opinion it’s harder to hear what higher sample rates get you, but the effect is there. The filters used to cut off frequencies above half the sample rate (the Nyquist frequency) can affect audio at lower frequencies. A lot of mastering engineers favor 88.2 kHz, because that’s a much higher frequency to set the filters and it’s mathematically much easier to convert to 44.1 from there (throw away every other sample, basically) than from 96 kHz. While the bar is being raised every day, sample rate conversion always does something audible to the music that should best be avoided if possible. Why have 48 and 96 kHz in the first place, then? Because those are common sample rates for video applications... music intended for DVD will sound best if set to that sample rate. Also, some early digital recording formats (like DAT) were set up to use 48 kHz to make it harder to copy audio from them to CD. DAT is gone, but the standard remains... When you’re done with your project and preparing it for CD, dithering down to 16 bits and 44.1 kHz should be your last step. And you should always save your original work as a high-resolution backup; someday a new format may come along, and it’d be too bad if you couldn’t take advantage of it because your music is only available on CD. Thanks for reading and I hope this helps.—MM Nothing gets past our readers! Props to you guys for once again putting out info that we hobby guys can use on our DIY music projects, this time on mastering. In fact, it got me so involved, I forgot that it was supposed to be your “April Fool’s” issue...until the “Fade Out” section! An informative, pertinent editorial on a real issue in the audio world, with some well-placed acronyms (Audio Performance & Recording Industry Levels/Friends Of Our Levels). APRIL FOOL, indeed— nice one!
Raymond Russell New York, NY

Dear editor: Thanks for alerting us of the continuing rights grab the Government is inflicting on the citizens of this country. I am forwarding that editorial to Sean, Glenn, Mark, Laura, Ann, and all the rest of the media patriots, including LIRPA, who will broadcast this info to the entire country so we can stop this totalitarianistic fascism in its tracks.
William Adams via email

PS. Could you send the address for the think tank Friends Of Our Levels?

Write to us at: Talkback, Recording Magazine, 5408 Idylwild Trail, Boulder, CO 80301 Or save stamps and send email to talkback@recordingmag.com
6
RECORDING JULY 2010

Recording - July 2010

Table of Contents for the Digital Edition of Recording - July 2010