Perfect Analog Tape Compression For Digital Recording^Popular

by Craig Anderton

If you've been trying to get true analog tape sound out of your MDM or hard disk recording (HDR)

system, you're gonna laugh when you read this. You can get a true analog sound more easily than you think, and no, it doesn't involve the messy process of synching a digital deck with an analog tape deck.

Before we reveal The Big Secret, here are some specifics to whet your appetite:

- True analog tape sound--not a simulation or "emulation"

- Comparatively inexpensive ($500-$700)

- Ability to provide sounds of different tape types

- Variable tape compression effects, from light to extreme

- Unlimited undo if you decide you didn't get it right the first time

- True stereo processing  ...and what would you pay for a signal processor that does all this? Well, read on.

The Secret Signal Processor

The secret ingredient is a three-head analog tape deck. This can be anything from a cassette deck to a spiffy Otari two-track. I use a Tascam Model 32, which you can pick up used for a reasonable price  (ever since DAT hit, the price of used two-track reel-to-reel decks has plummeted in musical circles, although I'm told reel-to-reel remains popular in the broadcast industry).

Fig. 1 shows how to patch the reel-to-reel deck into your system; here's the step-by-step procedure.

1.       Feed the already-recorded MDM or hard disk recording tracks you want to "analogify" into the reel-to-reel recorder inputs.

2.       Send the reel-to-reel outputs (monitor from the playback head) to two open MDM or HDR tracks. Set these to record.

3.       Load a reel of tape with the preferred sonic characteristics.

4.       Put the reel-to-reel into record mode, and roll tape.

5.       Start recording with the MDM or HDR. The tracks to be processed play into the reel-to-reel, while the open tracks record the processed sound.

6.       Do a trial run and adjust the reel-to-reel input level for the desired amount of crunch. Remember, you have to monitor from the playback head for this to work.

7.       After getting the sound you want, rewind to the beginning and transfer the tracks for reel--I mean, real.

We're not done yet, though, because the "crunched" signal will be delayed compared to the original,

non-crunched track. No problem: use the MDM or HDR track shift function to compensate. MDMs can delay but not advance tracks, so you have to delay the straight tracks to line up with the crunched tracks. With HDR, you have the option of advancing the crunched tracks in time rather than delaying all the straight tracks.

Monitor the crunched and non-crunched versions mixed together, then offset the original tracks until they line up with the crunched tracks (you'll hear a "flanging" sound as you get closer; go for the flanging "null" point). Now mute the original tracks, and you'll be left with pure analog tape sound. You need to figure out the appropriate offset only once, unless you change speeds on the reel-to-reel.

I did this with an ADAT/BRC combination and found that the right delay time for a Tascam Model 32

was about 75.3 ms at 15 IPS. I also discovered the wonders of flanging via track delay, but that's another story for another time.

One more tidbit: delay effects obtained by mixing the straight and processed sounds together can

sometimes sound very cool. I generally prefer having the straight sound hit late compared to the

processed sound.

Now, That Didn't Hurt At All

Not only is this technique simple, it allows your two-track to once more be a productive citizen of your

increasingly digitized studio. As a bonus, as long as you keep your original tracks, you can always go

back and re-crunch should you decide you crunched too much or too little (this is what I meant in the

beginning by "unlimited undo").

And that's all there is to it. Now you don't have to give up that analog tape sound, and best of all, you

won't erase some high frequencies every time you play it--and you can create as many digital safety

"clones" as you want.

Product opportunity alert: Some enterprising manufacturer could come up with a cut-down,

dumbed-down tape loop-based transport to do just this particular function. It may sound pretty wacky, but just think of it as an Echoplex for the '90s.

by Craig Anderton

My friend and occasional musical collaborator, Spencer Brewer, is sane in most respects except that he decided to build a studio. He was tired of the compromises and costs involved in going to commercial facilities, and after having several albums make the top 10 in the "adult contemporary" charts, decided it was time to take control over his own recording destiny.

Unlike many potential studiophiles, though, he spent the bucks to hire a professional studio designer--Dan Ryman, who had done several great-sounding rooms around California and is also a well-known sound designer/musician (e.g., the movie "The Color Purple") and concert mixer, including B.B. King, Pointer Sisters, Van Morrison, and Eddie Rabbit.

After visiting "Laughing Coyote" studios (located about 2.5 hours north of San Francisco in Redwood Valley), I was pretty impressed with the level of quality they had managed to obtain on a moderate budget. So, I twisted their arms and asked them to write up a dozen tips on things a potential studio constructor should do to insure a good-sounding room; here they are.

1. LOTSA GLUE IS GOOD FOR YOU

Using glue on every surface decreases the amount of sound bleeding through the construction materials, and ensures that the surfaces will never squeak. Use the highest grade of "tube squeezed" panel adhesive you can buy (preferably MD200 or 400 series glue). In building 1200 sq. ft. with 14' ceilings and 4 separate rooms, they used close to 78 cases.

Another 60 cases of clear silicone provided sealant between every wall, joist, rafter, floor, window, door, etc. Caulk all the joists first with a gun, then put on rubber gloves and spread the caulk in all surfaces that connect with the joists. It's major work, but remember--if air or water could get through, so can sound.

Finally, 30 cases of expanding foam were used to go into any spaces that were too large for caulk. Check out the different brands before you buy lots of cases; some work better than others. Do not get this on your hands or clothes when it's wet, because it will not come off!

2. GET IN TOUCH WITH YOUR INNER WALL

Cut 4" solid rigid foam to size and place it between every joist; caulk between the foam and the joists. Then, going from the outside of the wall inward, you have several layers (Fig. 1): 1/2" sheetrock, 1/2" soundboard (a very dense wood that's made out of pressed sawdust and glue), the joist/foam combination, 1/2" sheetrock, then 1/2" soundboard. All these sheets get screwed into the joists.

Leave a 3"-8" air gap, then add another mirror image "sandwich" (in other words, the layers go in the following order from air gap to wall surface: soundboard, sheetrock, joist/foam, soundboard, and sheetrock). The wall will end up around 16-26" thick, depending on whether you use 2 x 4" or 2 x 6" joists, and how much dead air space you have in the walls. This is one of the least expensive ways to add dead space in walls; more expensive methods put sand in between the air space, or other exotic approaches.

Fig. 1 - Wall construction detail.

3. DON'T BLOW A GASKET, USE TORCHDOWN

To further minimize leakage, use rubber Torchdown (other names are Dibitin or Meltdown) as a gasket for all the joists that are either touching the floor, an adjoining wall, or the ceiling. This is very dense rubber that is typically used on roofs (it's available at roofing supply house) and costs about $60 a roll. Glue the rubber on the wood first with the construction caulk gun because nails or tacks transfer sound into the wood.

4. DON'T BE A SQUARE

You don't want a square room, since this encourages standing waves that mess with your sound. Factor in a 2" to 3-1/2" lean from floor to ceiling (in other words, the top leans outward compared to the bottom) so that sound traveling across the room hits the reflective surface, then bounces up and dissipates in the ceiling texturing.

5. DOUBLE YOUR PLEASURE, DOUBLE YOUR GLASS

When building a see-through drum booth or isolation room, use two sliding glass doors separated by a dead air space of 4"-18". Hard, composite plastic sliding door systems resist sound migration better than metal. Since you'll end up with four panes of glass, don't use the normal tempered glass that is used in almost all sliding glass doors. Request clear glass, or your vision between rooms will be slightly "fuzzy" due to the layering of the glass.

6. A KINDER, GENTLER SURFACE

The surfacing on the walls is very important. Most home studios (and even some commercial ones) cover the walls with foam, curtains, egg cartons, etc. Using soft woods (pine, cedar, some kinds of birch, or for your most expensive option, spruce) at strategic locations for texturing provides a more sonically balanced room. You can't just throw up wood anywhere; you need to decide the type of balance you want between more reflective and more absorptive surfaces.  This is where a professional's viewpoint is essential. These wood accents are also visually pleasing, and bring warmth to the space (see Fig. 2).

Fig. 2 - Note the wooden accents on the far wall, just below where the ceiling meets the wall.

7. GO WITH THE (AIR) FLOW

One problem all studios face is the sound of air flow (from air conditioning or heating) into the rooms. A regular home air conditioning/heating unit will do unless you have a really huge amount of floor space; the trick is to use industrial size ducting. A normal duct is 8-10" in diameter. With 20-24" ducts, the air "falls" out of the vents instead of being pushed out, which reduces noise

dramatically.

Never split duct lines. Each line needs a clean line of travel back to the main unit, or bleed between rooms will occur. Also make sure the inner walls of the ducting have wool surfacing on them. If you're really a fanatic, wrap all your finished ducts with dense insulation for an extra quietness. Laughing Coyote ended up with 7 vents and 2 intake vents over 4 rooms; the entire cost was roughly $4,000 for heat and air conditioning. Finally, do not use vents that close outward. Almost all of the ducts you have should either be clear (no adjustments) or close inward so they block any air coming in. Otherwise the interaction between outgoing and incoming air can give a wheezing sound.

8. YOU GOT ME FLOATING (AND I'M FLOORED)

A floating floor is just that. Begin with a concrete slab (Spencer and Dan started with a 14' x 22' concrete slab in the main room). Next, buy a bunch of 3" x 3" rubber cushions with air holes in them, which are about 1-1/2" thick and available for around 10 cents each at flooring supply houses. Staple these to the bottom of 3/4" plywood, about a foot apart. Then lay the plywood on the concrete with the rubber surfaces face down (make sure you leave around a 1/2" - 1" gap between the flooring and the walls).

Next, lay down 3/4" particle board or chipboard in cross end patterns (in other words, if the plywood layer goes north-south, the particle board layer on top of it should go east-west). You have not used any nails, screws or glue up to this point. Finally, put 3/4" tongue and groove oak flooring on top of this using wood glue and an explosive flooring nail gun (which shoots 2" to 2-1/4" staples) into all three components. This flooring should also be in a cross pattern compared to the next layer down. The staples will travel through the various woods without connecting with the concrete. This results in a floor that floats above the concrete, which isolates sounds bouncing off it from the concrete.

For those who can't afford the $3.00-$4.50 a square foot for the flooring, there's another option. All large wholesale flooring companies sell what are called "shorts." These are the mill ends that are left over from red or white oak cuttings, and range in length from 14" to 28". Depending on where you live, "shorts" cost 90 cents to $2.00 a sq. ft. The red and white are generally mixed, but it makes for a very beautiful floor with great reflective qualities--and you save a bundle.

9. DEFINITELY NOT ROUGH TRADE

Run an ad in your local musician's magazine or local music store 2-3 months before your build begins, and offer to trade studio time for labor (carpentry, electrical, foaming, surfacing, demolition, cleanup, gopher, etc.). One can save thousands of dollars in labor costs this way and generate your first projects by trading for studio time, which generates publicity in your community.

10. DOUBLE YOUR PLEASURE, DOUBLE YOUR BUDGET

Whatever your budget and time constraints are, double it. Spencer figured with all the deals he was getting, the studio would be done for $35,000-$40,000 in 3 months; due to a variety of unplanned additions designed to enhance the studio even further, it ended up taking almost 6 months and costing $85,000. Still, it looks and sounds like a $200,000 room, mostly because of the deals cut on lumber, labor trades, and Spencer serving as his own gopher (which took 3-5 hours a day regularly). He recommends that the owner be the gopher since you need to build relationships with those in the area who have connections to deals.

11. BUILDING INSTANT KARMA FOR BUILDING

Spencer plays a lot of benefit concerts in his area for non-profit groups, which ended up giving him connections he would not have had otherwise. For example, he traded a complete set of his CDs (14 at the time) to a local lumber yard owner for a 100 year old 6" x 12" x 40' redwood beam for the main ceiling.

12. CONSTRUCTION REQUIRES PRE-PRODUCTION

When you work with a studio designer, get upfront advice and do lots of work on paper. A program like Virtus Walkthrough (available for both Mac and PC) lets you model the space in advance of construction and literally "walk" through it on your computer, which can be very helpful. As to the designer, listen to previously designed rooms and make sure that the response is true. Just remember, the one thing all your recordings will have in common is the room, so don't skimp on it--even if your needs are not quite as complex as the ones described above. You can always upgrade your gear, but a good room is forever.

by Craig Anderton

Too many people adjust equalization with their eyes, not their ears. Example: after doing a mix recently, I noticed the client writing down all the EQ settings I had made. When I asked why, he said it was because he liked the EQ and wanted to use the same settings on these instruments in future mixes.

Wrong! EQ is a part of the mixing process; just as levels, panning, and reverb are different for each mix, EQ should be custom tailored for each mix as well. But to do that, you need to understand how to find the magic EQ frequencies for particular types of musical material, as well as what tool to use for what application.

Simply stated, there are three main applications for EQ:

- Problem-solving

- Emphasizing or de-emphasizing an instrument in a mix

- Altering a sound's personality

Each application requires specialized techniques and approaches.

PROBLEM-SOLVING

EQ can fix some obvious problems. Examples: Slicing a sharp notch at 60 Hz (50 Hz in Europe) can knock hum out of a signal; trimming the high frequencies can remove hiss. Generally, problems occur in specific frequency ranges, which makes the parametric type of equalizer ideal for problem-solving (see sidebar, "Parametric vs. Graphic").

Another common problem is an instrument with a resonance or peak that interferes with other instruments, or causes level-setting difficulties. Following is a procedure that takes care of this situation.

The Case of the Classical Guitar: In 1977 I produced an album for Tomato Records by classical guitarist Linda Cohen (Angel Alley, recently re-released on CD). She had a beautiful instrument with a full, rich sound that projected very well on stage, thanks to a strong body resonance in the lower midrange that caused a major level peak. However, recording was a different matter from playing live. If levels were set so the peaky, low frequency notes didn't overload the tape, the higher guitar notes sounded weak by comparison.

Although compression/limiting was always an option, it altered the guitar's attack; while this effect might have gotten lost in an ensemble, it stuck out with a solo instrument. A more natural-sounding answer was to use EQ to apply a frequency cut equal and opposite to the natural boost, thus leveling out the response. But there's a trick to finding problem frequencies so you can alter them; the following works like a charm.

1.       Turn down the monitor volume: things might get nasty and distorted during the following steps.

2.       Set the EQ for lots of boost (10-12 dB) and fairly narrow bandwidth (around a quarter-octave or so).

3.       As the instrument plays, slowly sweep the frequency control. Any peaks will jump out due to the boosting and narrow bandwidth. Some peaks may even distort.

4.       Find the loudest peak and cut the amplitude until the peak falls into balance with the rest of the instrument sound. You may need to widen the bandwidth a bit if the peak is broad, or use narrow bandwidth for single-frequency problems such as hum.

This technique of boost/find the peak/cut can help remove midrange "honking," strident resonances in wind instruments, and much more. Of course, sometimes you want to preserve these resonances so the instrument stands out, but many times applying EQ to reduce peaks allows instruments to sit more gracefully in the track.

Digital workstation EQ, as found in hard disk recording systems, can be particularly effective due to its precision. In one of my more unusual projects, I needed to remove boat motor noise from some whale samples. Motor noise is not broadband, but exists at multiple frequencies. Applying several extremely sharp and narrow notches at different frequencies took out each component of the noise, one layer at a time, until the motor noise was completely gone.

This type of problem-solving also underscores a key principle of EQ: 'tis better to cut than boost. Boosting uses up headroom; cutting opens up headroom. In the example of solving the classical guitar resonance problem, cutting the peak allowed for bringing up the overall gain to print a lot more overall level on tape.

EMPHASIZING INSTRUMENTS

The same technique of finding and cutting specific frequencies can also eliminate "fighting" between competing instruments. For example, while mixing a recent Spencer Brewer track for Narada records, there were two woodwind parts with resonant peaks around the same frequency. When playing en ensemble they would load up that part of the frequency spectrum, which also made them difficult to differentiate. Here's a way to work around this:

1.       Find, then reduce, the peak on one of the instruments to create a more even sound.

2.       Note the amount of cut and bandwidth that was applied to reduce the peak. 

3.       Using a second stage of EQ, apply a roughly equal and opposite boost at either a slightly higher or slightly lower frequency than the natural peak.

Both instruments will now sound very articulated, and because each peaks in a different part of the spectrum, they will tend not to step on each other.

NEW SONIC PERSONALITIES

EQ can also change a sound's character--for example, turn a brash rock piano sound into something more classical. This type of application requires relatively gentle EQ, possibly at several different points in the audio spectrum; a graphic equalizer works well. Parametric EQs may not have enough bands to affect all the desired sections of the audio spectrum.

Musicians often summarize an instrument's character with various subjective terms. Fig. 1 correlates these terms to various parts of the frequency spectrum (this is, of course, a very subjective interpretation).

Fig. 1: Frequency ranges translated into musical terms.

For example, to add warmth, apply a gentle boost (3 dB or so) somewhere in the 200-500 Hz range. However, as in the previous case, remember that if possible, cutting is preferable to boosting--for example, if you need more brightness and bottom, try cutting the midrange rather than boosting the high and low ends (Fig. 2).

Fig. 2: The upper graphic equalizer boosts the high and low frequencies; the lower graphic cuts the midrange to accomplish the same effect, but with less likelihood of distortion.

OTHER EQ TIPS

Problem-solving and character-altering EQ should be applied early on in the mixing process since they will influence how the mix develops. But wait to apply most EQ until the process of setting levels begins; remember, EQ is all about changing levels--albeit in specific frequency ranges. Any EQ changes you make will alter the overall balance of instruments.

Another reason for waiting a bit is that instruments EQ'ed in isolation to sound great may not sound all that wonderful when combined. If every track is equalized to leap out at you, there's no room left for a track to "breathe." Also, you will probably want to alter EQ on some instruments so that they take on more supportive roles. For example, during vocals consider cutting the midrange a bit on supporting instruments (e.g., rhythm guitar) to open up more space in the audio spectrum for vocals.

Finally, remember that EQ often works best when applied subtly. Even one or two dB of change can make a significant difference. However, inexperienced engineers often do something such as increase the bass too much, which makes the sound too muddy, so they increase the treble, and now the midrange sounds weak, so that gets turned up...you get the idea. One of your best "reality checks" is an equalizer's bypass switch. Use it often to make sure you haven't lost control of the original sound. (In a perfect world, all mixers would have bypassable equalization.)

So there you have it. Go forth and tweak those tones!

Beginner's Corner: Graphic vs. Parametric EQ

A graphic equalizer divides the audio spectrum into numerous narrow bands (generally up to 32), each with an associated vertical slider that boosts or cuts the amplitude at that specific frequency. The slider's physical positions create a rough approximation of the frequency response curve, hence the term "graphic" equalizer. New digital versions dispense with the sliders in favor of buttons and readouts, but are still called graphic equalizers.

A parametric equalizer has a lesser number of more versatile bands where you can change

not just the degree of boost or cut, but also the frequency at which this occurs and the response bandwidth. "Pseudo-parametric" equalizers omit the bandwidth control, which can seriously hamper the experienced EQ aficionado. Arguably, manufacturers err on the side of too narrow a fixed bandwidth, which makes it difficult to do subtle changes.

by Craig Anderton

Auxiliary busses aren't too exciting, right? You send some signal to a reverb, then bring the reverb back into the effects returns. Then you add some reverb. Big deal.

Ah, but those aux busses can do a whole lot more--exploit them, and you can make your mixes just about jump out of the speakers. The secret is to use not only reverb, but a variety of signal  processors that can warp and bend your sound in wonderful ways. We'll present those tips in a minute, but first, a word of advice if you don't have enough aux busses to go around.

Many reverb devices claim to be stereo, but what they really do is sum the two inputs together and

process the resulting mono signal into stereo (Fig. 1).

Fig. 1

Therefore, there's no need for a channel to be able to choose between reverb inputs since the signals end up in the same place anyway. Unless your reverb is true stereo it will sound just as good if you use one aux send per channel, which frees up the other send for something else.

And now, two cautions:

Whatever signal processor you use should be set for processed sound only (no dry signal). The

channel faders themselves provide the dry signal.

It's important that the processor signal be in-phase. Otherwise, the more processed signal you

bring up, the greater the odds of phase cancellation messing with your sound.

Now let's look at aux bus options other than reverb.

EXCITER

Most of these insert between mixer and master deck, and therefore they process the entire stereo master signal. However some of the signal sources being mixed, especially samples, may already have been "excited" somewhere else along the line; when processed again, they become strident and harsh. Another problem: if a signal is noisy, the exciter will emphasize the noise (boo, hiss--but mostly hiss).

Sending some channels but not others to an exciter opens up lots of possibilities. For example, with a combination of electronic and acoustic instruments, adding exciter to the acoustic tracks helps them hold their own with the brighter electronic timbres, thus creating a better overall balance. Once you've used an exciter in this somewhat more refined fashion, you'll have a hard time going back to just slapping the thing onto two tracks.

DISTORTION

Actually, this is a relative of the exciter family, but provides a "thicker," rather than "airier," timbre. I

favor solid-state tube emulators (such as the SansAmp PSA-1) for this application because they tend to give a wider variety of sounds than tubes, and generally provide programmability for more repeatable settings.

The trick here for me has been to pick a couple of instruments to be emphasized, and bring up a little

distortion behind them in the mix. Favorite candidates for this include vocals and drums. I believe this

also helps bring back some of the distortion-related aspects of saturating analog magnetic tape, which we associate with "pushing" the sound.

Using distortion for obvious effects a la NIN is one thing; getting a subtle, warming effect takes some

work with EQ. Trimming off highs before feeding the distortion gives a smoother, rounder tone, and lets you bring up the distortion higher before it becomes noticeably ugly. Cutting bass is another matter altogether. This produces an extremely brittle, bright high end that is effective if mixed way-in the background--treat it like an exciter with an attitude problem, and you should get the levels about right.  Experimenting with EQ after the distortion, in particular slightly boosting the lower midrange, is also worth trying.

My main hookup for messing with distortion uses a Rane MPE-14 stereo graphic EQ patched before and after the SansAmp (Fig. 2).

Fig. 2

I use this effect in mono since I'm not interesting in splattering distortion across the stereo field, but just give a little mono buildup for the selected sounds. Stereo is cool too, but I'm not sure it's cool enough to justify dealing with another distortion/graphic EQ pair.

VOCODER

This technique is remarkably cool for dance music. I'm sure some EQ reader will hook this up and cut a giant hit (if you do, send me a copy!).

For those who don't know what a vocoder is, it makes "talking instrument" sounds--sort of like the

talkbox effect that was so popular in the 70s (Peter Frampton, Stevie Wonder, Joe Walsh). In a nutshell, you plug a mic into one input (the modulator), and plug the signal you want to have "talk" into the  carrier input. As you speak, your voice puts out more or less energy in various frequency bands; the carrier goes through filters that let through more or less energy depending on what your voice is doing.  This impresses the vocal effect onto the carrier (for detailed information on vocoders, see my book "Multieffects for Musicians," published by AMSCO).

Anyway, there's no law that says you have to use a vocoder only with voice. To translate this to

mixer-speak (Fig. 3), one aux send feeds the carrier, and another aux send, the modulator. This allows any channel to serve as either modulator or carrier (you can also have a signal modulate itself, which is a whole other story).

I like to use the drum track as the modulator, and rhythm guitar parts or background keyboards as the carriers. This superimposes a highly rhythmic effect on the guitars/keyboards.

Fig. 3.

SHORT DELAYS

Short delays are a useful supplement to conventional reverb. Using a combination of room reverb and short delays can build a bigger ambiance than just adding more or less "standard" reverb, as the tracks with more short delays generate the equivalent of more early reflections.

THE MAGIC BUS

Those who gravitate toward using less tracks but having each musical part convey more substance will find this approach particularly appealing, as you fill out the sound not by adding more notes or

instruments, but by "modeling" a more interesting listening environment. If your mixer takes a few

detours into aux bus-land before hitting the stereo master deck, you can often end up with a much

sweeter, and more attention-getting, mix.

by Craig Anderton

Timing is everything, and that's especially true with music. Yet mathematically perfect timing is most certainly not everything, otherwise drum machines would have replaced drummers a long time ago. Good drummers enhance music by playing with the time--subtly speeding up or slowing down to change a tune's "feel," and leading or lagging specific beats to push a tune or make it lay back a bit more in the groove.

Often, these time changes ahead of or behind the beat are very small; even a few milliseconds (ms) can make a difference. This is surprising, since sound itself moves at about 1 foot per second, so a 6 ms change theoretically affects a track about as much as moving an amp 2 yards further behind the drummer. Yet when you conduct timing shift experiments, it becomes obvious that even very small timing differences can change a tune's groove when you hear these changes in comparison to a relatively steady beat.

Musicians and engineers often forget about the importance of timing changes and quantize everything, which is the quickest way to suck the life out of a piece of music. Fortunately, we can use other aspects of sequencers (and more recent drum machines) to put the feel back in to sequenced music.

A SHIFTY TRACK GIVES AN HONEST FEEL

Many sequencers provide timing randomization options to help give a more human-sounding track. Randomization is great if you want to simulate the effect of a drummer who's had too many beers; however, for a groovacious rhythm part, shift timings the way a drummer would. Human drummers add variations in a mostly non-random way--often subconsciously, so these changes tap directly into the source of the drummer's "feel."

Drummers often hit some drums slightly ahead of, or behind, the beat to give certain effects. For example, jazz drummers tend to hit a ride cymbal's bell a bit ahead of the beat to "push" a song. Rock drummers frequently hit the snare behind the beat (listen to any Led Zeppelin album) to give a "big" sound. Of course, the sound isn't really bigger; but our brain interprets slight delays as indicating a big space, since we know that in a big space, sound travels a while through the air before it reaches us.

A sequencer or drum machine's track shift (or track offset) function, which can move a track back and forth in increments of single clock pulses, is your first line of defense against mechanical grooves. Keep the kick drum on the beat as a reference, and use track shifting to change the timing of the snare, toms, and percussion by a few milliseconds. Here are some other track timing tricks.

- For techno, dance, and acid jazz tunes try moving any double-time percussion parts (shaker, tambourine, etc.) a little bit ahead of the beat to give a "faster" feel.

- Sometimes it pays to shift individual notes rather than an entire track. With tom fills, delay each subsequent note of the fill a bit more (e.g., the first note of the fill is on the beat, the second note approximately 2 ms after the beat, the third note 4-5 ms after the beat, the fourth note 6-8 ms after the beat, and so on until the last note ends up about 20 ms behind the beat). This can make a tom fill sound gigantic.

- If two percussion sounds often hit on the same beat in a rhythm pattern, try sliding one part ahead or behind the beat by a small amount (a few ms) to keep the parts from interfering with each other.

- If some drums fight with melodic parts (e.g., the kick drum and bass mosh together), slightly advance the part you want to emphasize in the mix. It will grab the ear's attention just before the beat, therefore bringing more attention to itself.

- Track shifting does not apply only to drum parts. Suppose there are two fairly staccato harmony lines in a tune. If you advance one by 5 ms and delay the other by 5 ms, the two parts will become more separate and distinct instead of sounding like one combined part. If the parts are panned oppositely in the stereo field, the field will appear even wider.

- Hitting a crash cymbal a bit ahead of the beat makes it really stand out. Moving it behind the beat meshes it more with the track.

QUANTIZATION--TOOL OF SATAN?

Remember, machines don't kill music, people do--and quantization is one of the main weapons. Although quantization has its place, it's a very artificial process because no drummer plays with crystal-controlled precision.

Fortunately, sequencers usually let you quantize by a certain percentage (usually called "quantize strength" or "intensity"). In other words, 100% quantization moves a note exactly to the nearest beat, but 50% quantization moves it halfway closer to the beat. I quantize the kick to 100% and all other drum tracks to somewhere between 50% and 80%. The result is a track that sounds rhythmically correct, but retains most of a performance's "feel."

TERRIFIC TEMPO TRACK TIMING TWEAKS

People generally set the tempo in a sequencer to the desired beat, then just lit it sit there. That's not the way real music works; in a fascinating study, Ray Williams and Ernest Cholakis (of DNA Groove Templates) compared the tempo tracks of two classical pianists playing Moonlight Sonata, and plotted out the tempo changes. The results were anything but a constant tempo--the changes looked like a relief map of the Alps.

Even though pop music doesn't change tempo as much as classical pieces, real drummers insert subtle tempo changes, inserted over several measures or just in selected parts of individual measures, to build anticipation and change moods.  Fortunately, most sequencers let you change the tempo track throughout a song; once you start working with this technique, you'll find it an essential part of the production process. Here are some examples of track shifting.

- To boost a song's energy level, increase tempo slightly (by 1 or 2 beats per minute). This is the timing equivalent of modulating pitch upward by a semitone; both increase excitement. Decreasing tempo has the reverse effect.

Tempo shifts are useful when transitioning between song sections (verse to chorus, chorus to instrumental, etc.) as well as within a particular section (such as upping the tempo for the last two measures of a solo).

- Change tempo a little bit before the first beat of the measure you want to change. For example, if you're going from verse to chorus, increase the tempo halfway through the measure prior to the chorus. This creates a smoother lead-in than having the tempo change coincide with a measure change.

- For really dramatic effects, ritard the tempo over the course of a measure (e.g., one BPM or less lower on each beat) then return to the original tempo. Having a drum roll during the ritard creates a particularly effective transition.

NOT JUST TIMING--TIMBRE TOO

Along with timing, here are a few suggestions on better drum parts through better timbres.

- To simulate the fact that two consecutive drum hits never have exactly the same timbre, assign the same drum sound to two different notes. For example, with an Alesis D4, copy the snare sound to a different MIDI note, then detune the alternate sound by a very small amount. There will be a subtle, but noticeable, timbral difference between the two snare sounds. Shift every other note or so of your snare part to trigger the second snare, and you'll have a much livelier part.

- With sampled drum sounds, route some velocity modulation to pitch so that high-velocity hits are slightly higher-pitched. This gives the feeling of the drum skin being stretched tighter; set the pitch modulation amount so that the change is not really audible except when compared to the non-bent sound.

- Velocity also works well when routed to filter cutoff so that harder hits give a brighter sound.

- Set a drum sample's start point several milliseconds after the start of the sound, and use velocity to push the start point closer to the sample's beginning. Hard hits will give a louder, more forceful attack; the effect can be much more convincing than velocity-switching between different samples (although of course, this has its uses too.)

Before signing off, I'd like to thank Michael Stewart (Digidesign) and Marius Perron (Jeanius Electronics) for sharing their insights about timing with me; several parts of this article owe a lot to their research. The more you get into the complex interplay between timing and sound, the more interesting it gets.

by Craig Anderton

Mastering has typically been the weak link in the project studio. Sure, you can mix down to DAT--but how about assembling those mixes into a smooth, well-integrated recording, adding P and Q codes, and transferring over to a format like Sony 1630? Although it's still difficult to do everything without going to a real mastering facility, you can nonetheless do several of the most important parts of the mastering process in today's project studio.

Once again, hard disk and 8-track digital tape technology has come to the rescue. If you mix to DAT and have something like an Alesis ADAT or Tascam DA-88, you already have a great pre-mastering machine and may not know it. If you also have a hard disk recording/processing system a la Sound Tools, you're even further ahead. But first, let's talk a bit about why mastering is so important.

MASTERING BASICS

Proper mastering can make a marginal recording acceptable, or a good recording great. Often the difference between what comes out of a modest project studio and a multimillion dollar facility is not in the recording, but in the mastering.  Good mastering engineers are rare, because they need to make flawless aesthetic decisions as well as have total command over signal processing technology.

Here are some of the procedures used during the mastering process that you can do in a project studio:

- Balance levels between different cuts, or even within different sections of the same cut.

- Apply overall EQ to add "sheen," or to compensate for problems (e.g., remove bass caused by bad monitors that led to a bass-heavy mix).

- Make a tune more "radio-ready" by adding compression or limiting to allow a higher average signal level.

- Crossfade between cuts.

- Add processing, such as a hint of reverb to tunes that seem too dry.

- Create fade outs and fade ins.

- De-noise noisy sections, either through gain-riding, software programs, or single-ended noise reduction systems.

MASTERING THE HARD (DISK) WAY

Quite a few people use hard disk digital audio systems for mastering. The basic idea is to bounce your tracks over to the hard disk system (digitally, if you have a DAT with digital I/O), then apply digital EQ, limiting, gain changes, etc. You can then assemble a playlist to try out different song orders, and when everything is as desired, transfer the results back to DAT for your final master tape. However, there are some limitations involved with using only hard disk systems.

- You sacrifice real time control, which is very important with mastering. For example, to process a piece of music, you usually have to set up the parameters, then wait while the computer crunches its numbers to do the processing. It's cumbersome with budget hard disk systems to do something like increase the treble a few dB over several measures, then pull it back a bit later.

- There are device-specific limitations. For example, with Sound Tools crossfades are done in RAM--and you know how digital audio gobbles up memory.

- It ain't a totally digital world (at least not yet). I have a bunch of nifty analog processors, from the obsolete-but-fun EXR Projector to the new Aphex model 104, that are ideal for mastering and have no equivalent function in any hard disk system I've seen.

THE TAPE TEST

Adding a digital multitrack to the equation can overcome these limitations; the only tradeoff is a theoretical loss of quality if you bounce using the analog inputs (of course, if you can bounce digitally, this isn't an issue). In practice, though, the difference in quality may not be noticeable. There may even be a subjective improvement. To find out if your system is up to the task, try this experiment:

1.       Mix some tracks to DAT.

2.       Bounce the DAT over to two tracks of your digital deck.

3.       Bounce the digital deck tracks back to DAT.

4.       Compare the original DAT sound with the one that was bounced to the digital deck and back. If the bounced version sounds okay to you, you're ready to master.

SCALPEL, SUTURE, AUDIO: START OPERATING

The following summarizes how I do mastering with a Tascam DA-30 DAT, Sound Tools, and one Alesis ADAT. You shouldn't have any trouble modifying this procedure to fit your specific setup; it's the principles that are important.

1.       Record mixes of all your tunes to DAT.

2.       Bounce the DAT mixes digitally to Sound Tools, then use it for what it does best: normalization to make sure you're using the maximum available headroom, peak limiting to let you get a bit higher average level, and overall EQ changes. Sometimes these interact; for example, you might normalize the tune, limit the peaks to open up more headroom, then normalize again. Or, cutting some of the bass might lower some peaks, allowing for re-normalization.

3.       Bounce the processed tune digitally back to DAT, but don't go over your original mix so you have the original as backup.

4.       Repeat steps 1-3 until all the tunes are done.

5.       Next, figure out the optimum order for the tunes. Do this by recording them all into Sound Tools and trying out several play lists until you get the right order.

6.       Now it's time to assemble. Patch your DAT outs to ADAT tracks 1 and 2 (now's your big chance to include any analog processing, if appropriate), and record the first tune into ADAT.

7.       Record the next tune on tracks 3 and 4. Notice how easy it is to do crossfades with this technique; just start recording the second tune sometime before the first tune ends.

8.       Record the third tune into tracks 1 and 2, and keep alternating tunes between tracks 1/2 and 3/4 until all the tunes are in ADAT.

9.       Note that you still have four tracks left over. You can use these to insert additional effects or transitions (great for dance mixes), or add time code for reasons we'll get into next.

10.   Now mix the ADAT tracks back down to DAT to end up with a final 2-track DAT master. If you need to make any volume tweaks, you can do so manually or by synching automation to time code recorded on an ADAT track (or generated by the JL Cooper box or Alesis BRC).

11.   If needed, you can now bounce the DAT back to Sound Tools to create additional masters for different purposes. For example, if the piece is going to be duplicated on cassette as well as CD, I often add just a tiny bit more compression and "exciter" treble enhancement to compensate for losses in the duplicating process.

THERE'S NO PLACE LIKE...

Three of the best reasons to master at your own studio are the same reasons for recording there: you can learn a lot without spending beaucoup bucks, you have more control over the final product, and you can do things over and over (and over) again until you get the sound just the way you want it. And with today's mastering tools, you have extra creative options that simply didn't exist a few years ago. Try using digital multitrack for mastering--you'll see what I mean.

by Craig Anderton

The Achilles heel of digital audio is storage--at 10.5 Megabytes per minute of stereo recording, you can chew through a hard disk in no time. But recording the signal in the first place is not as much of an issue as backing it up. Even a gigabyte hard drive probably won't provide enough storage to keep all your data for current projects, as well as backup data for previous projects.

Consider the options. Removable cartridges? Nice, but can be costly as your storage needs multiply. Magneto-optical drives are a big step up for backup, but they're still rather pricey, and the technology isn't really fast enough for multitrack hard disk recording.

DAT backup? I don't know about you, but DAT--which was never intended for pros--seems like a somewhat tentative backup medium. Those little teeny tracks on those little teeny fragile cassettes does not inspire feelings of tremendous confidence. Still, DAT backup does offer the advantage of being cheap ($10 per DAT) and storing humongous amounts of data.

And that's what I used until I learned of a special offer for ADAT owners: a removable cartridge drive for $895 that stores up to 1.6 Gigabytes of data  (partitioned into four 400 Meg sectors) on an $11 cartridge that's much more robust than a DAT. This offer is also available to DA-88 owners (who will

obtain even more storage--over 4 Megs partitioned in four 1.1 Gigabyte sectors), as well as other digital audio recorders capable of recording and playing back back through an AES/EBU or SPDIF interface. Sound good to you?

TALE OF THE TAPE BACKUP

The workhorse for inexpensive tape backup is the digital multitrack recorder itself, but the secret ingredient is an AES/EBU compatible interface. The "$895" I referred to above is actually the cost of an AI-1, the AES/EBU interface/sample rate converter for ADAT. This is what enables you to dump

digitized audio data directly onto the multitrack tape. For DA-88 owners, the Tascam IF88AE digital interface provides an equivalent function.

If a device can send and receive audio as AES/EBU digital data, that data can be converted into the multitrack's proprietary digital format by the AES/EBU converter, and stored on tape. 

There are several advantages to backing up on digital multitrack tape:

·        Inexpensive, universal storage medium. Rather than having some audio data on Syquests, some on DAT, and some on floppies, everything can be saved onto a common format.

·        It's easy to "clone" tapes and make safeties if you have access to a second digital multitrack.

·        Pieces of data can be referenced to the recorder's time reader, so if you have an autolocator, you can simply enter the time where the data resides, then retrieve it. (With the ADAT BRC, you can name location points and store them as a "song." This makes it real easy to find and recall data.)

·        If you're backing up mono audio, you can store twice as much as mentioned above (3.2 Gigabytes for ADAT, and 8.8 Gigabytes for the DA-88).

·        It's easier to exchange audio data with other studios since it's more likely they'll have a digital audio multitrack recorder than something like a 600 Meg optical drive.

However, you don't get something for nothing. The main tradeoff is that backups generally have to be done in real time, and restoring is a real time process as well. Also, tape falls somewhere in the middle of the robustness scale; it's probably better than some of the more fragile removables, but not as good as a magneto-optical cartridge. The biggest disadvantage is that you can't back up other types of computer files, since only digital audio is eligible.

DETAILS, DETAILS

Because of the "partitioning" effect of having several different tracks, you'll have to decide which files should go to which tracks, and keep accurate records of what data is located at what times on what tracks (I made up a form for entering this data--it definitely helps). Still, ADAT's 400 Meg and the DA-88's 1.1 Meg partitions are not exactly chintzy, and you'll probably encounter few projects that need more contiguous storage space than that.

Backing up DAT is simple: just plug the digital out into the ADAT or DA-88's digital I/O input, put the tracks you want into record, and play the DAT. Samples are a different matter. For my Mac-based system, I use Digidesign's DAT I/O hardware and DATa software. Once the sample is imported into

sample editing software (Alchemy, Infinity, or Sound Designer), it's saved to hard disk. DATa can save and restore either individual files or folders of individual files.

A few fine points:

·        If you're using DATa, save all your files as Sound Designer II files before backing them up. DATa will also back up AIFF files, but the process takes longer.

·        DATa writes a header than includes loop points and other file characteristics as well as the audio itself. However, Sound Designer II files are recoverable as standard audio should this become corrupted.

·        I try to keep everything in my studio at 44.1 kHz, which initially presented a problem because of ADAT's default 48 kHz sampling rate. However, setting ADAT's pitch control to -1.47 provides a 44.1 kHz sampling rate (as confirmed by the pitch control readout), so you can do a straight 44.1-to-44.1 transfer for backup. (The AI-1 does allow for format conversion, so you can bounce 44.1 to 48 or vice-versa if need be.)

·        Store your tapes properly (a cool, dry environment is optimum).

GET BACKUP TO WHERE YOU ONCE BELONGED

Stepping back for the bigger perspective, although multitrack tape recorders provide very inexpensive backup, this is not necessarily the most convenient method. I use a 100 MB ZIP drive as a "holding tank" for current work to avoid the time required to do a tape-based backup and restore onto my main hard disk.

Unfortunately, a 100 MB disk is really only good for backing up fairly short projects. Eventually, I'd like to replace it with a mondo optical drive when their prices come down a bit. Even then, though, I expect I'll continue to make backups on tape. Optical cartridges are still much more expensive than

something like an S-VHS tape, and for archival files that may or may not get used again, tape seems like a much more cost-effective way to go. It certainly has worked well for me.

by Craig Anderton

The reverb-drenched sound of the '60s and '70s is well behind us, as is the chorused gauziness of the '80s. The sound of the 90s is high-definition and in your face; you hear less and less reverb on records, whether you're talking Red Hot Chili Peppers or cutting-edge dance music.

You might think going direct is one way to get this effect, but we're used to instruments having some "air," both from resonances within the instrument itself and from the room in which it is played. Listen to a drum machine going direct into the board: yes, the sound is clean--but there's also a certain deadness. The stereo is too wide; drums become individual points of sound instead of being part of a cohesive, unified kit. A synthesizer or sampled keyboard--in fact, any electric or electronic sound source--suffers from similar problems when going direct.

Some recording engineers even pump electronic sounds through speakers and then mic them (not at all a bad idea, by the way), but there's a more predictable and compact way to give your electronic sounds some air: model a room.

AMBIENCE FOR THE '90s

While "modelling" is a big buzzword these days, the concept has been around for a while. Any electronic reverb is essentially modelling what happens when sound waves run around loose in a room.

Recording in a very tight, sparse, hard "box" of a room is one way to get that in-your-face sound. Back in the early days of digital delay, one technique to simulate this kind of ambience was to put several delays (with delay times of 1 to 10 ms or so) in parallel. Mixing these delays well in the background creates the "comb filtering" effects associated with typical small rooms.