Primed for Mixdown

Image placeholder title
Image placeholder title

FIG. 1: A bass waveform recorded without clipping (a) and clipped (b)

With a good DAW, mic and mic pre, you can record master-quality tracks. A master-quality mix, however, often requires more expensive gear and the specialized skill set of a mix engineer. For this reason, many producers record tracks themselves and then send them to a mix engineer.

As a producer/arranger and a mixer, I've had engineers grumble about technical or organizational issues with the tracks I've sent them, and I've done my share of grumbling about tracks sent to me. In this story, I'll offer tips and advice that will lessen the grumbling and make your finished product sound better. Even if you're doing the mixing yourself, these tracking and editing tips will make your job a lot easier.

On the Level

Digital distortion, or clipping, is a big problem. It adds noise to the mix, which reduces its clarity. It makes the tracks that are distorted sound small and undefined. And there's nothing that can be done to fix it. When clipping occurs, any part of the waveform that goes over the limit of the recording equipment is chopped off, leaving a signal that looks like a mesa rather than a mountain (see Figs. 1a and 1b). If the clipping is severe, it's audible as an unpleasant noise. If it's less severe, the track loses audio quality. There is never a good reason to overload a digital recording at the tracking stage. If you want digital distortion in your music, you can always add it later simply by overloading any stage of the signal chain.

Image placeholder title

FIG. 1: A bass waveform recorded without clipping (a) and clipped (b)

The best way to avoid distortion is to always use 24-bit audio in your recordings rather than 16-bit, and record at more moderate levels. A 24-bit recording uses 50-percent more disc space, but gives you 256 times the headroom of 16-bit recording (see Online Bonus Material“Do the Math”).

It's particularly tempting to record vocals hot because singers want to hear themselves clearly in the headphones while recording their parts, and the easiest way to satisfy them is to simply turn up the input level on the mic pre. But it's much better to keep the level on the mic pre down and boost the singer's monitor levels. If that's still not enough, lower everything else in your mix and raise the headphone levels. If that's still insufficient, you need a more powerful headphone amplifier.

It's also tempting to record drums and synths with a little clipping because they can sound edgier that way. Again, that edginess can be added later. Clean tracks give the mix engineer a lot more flexibility. So if you see clip indicators anywhere in your tracks, go back and record them again at a lower level. If you are recording live performances, or if you know that you won't be able to re-record a clipped take, keep your levels conservatively low, -6 dB or -12 dB, so you don't come close to going over.

Click Click

A mix engineer will generally brighten and compress tracks, and both of these actions raise the level of clicks and pops. Digital clicking, particularly, will cut right through a mix. Before you send a track to a mix engineer, listen to each of the individual tracks, soloed, and get rid of clicks and pops. Most often they will be the result of bad punches or edits. Move the edit points around, and use crossfades until the clicks and pops disappear (see sidebar “Eradicate Clicks and Pops” on p. 52).

Image placeholder title

FIG. 2: Vocal tracks where the parts in different sections of the song were recorded on separate tracks (a).The same song with sections grouped onto single tracks for each vocal part (b).

Virtual instruments can also make digital clicks if the computer's CPU is overtaxed. If you hear clicks on a track that was recorded from such an instrument, disable all other virtual instruments and re-record the track.

A singer's mouth noises are difficult to eliminate. They can sound like digital clicks and, in a quiet song, they too will cut right through the mix. If a singer is making a lot of audible mouth noise, I sometimes ask them to put mineral oil on their lips. If that doesn't work, a judicious use of audio restoration plug-ins, or the restoration features built into your audio editor, can help eliminate the clicks after the fact.

Image placeholder title

FIG. 2: Vocal tracks where the parts in different sections of the song were recorded on separate tracks (a).The same song with sections grouped onto single tracks for each vocal part (b).

Organizing Principles

As a general rule, and to keep confusion to a minimum, you want to combine tracks that belong together onto a single track (see Figs. 2a and 2b). For instance, I'll sometimes get tunes that have the verse backup vocals on one set of eight tracks, the chorus backups on another set of eight and the bridge backups on yet another eight. If there are no overlaps in the parts and you don't want a different sound for each section, these tracks should be combined into one master set of eight.

On the other hand, if you programmed or recorded the strings in separate instrument sections (vc, vla, vl2 and vl1, for example), keep those separated to give the mix engineer maximum control. Drums should also be split out as much as possible. If you are using a drum plug-in such as FXpansion BFD and you want a realistic drum sound, print all the outputs on separate tracks — including the top and bottom snare mics, inside and outside kick mic, room, overhead and ambience. Mix engineers who mix live drums treat these tracks differently to get their sound.

Tracks that contain many separate audio files, punches and crossfades can sometimes get corrupted as you send your song file over the Internet. It's best to consolidate (or merge) those tracks, but only after you've carefully checked them for clicks, pops, funny breaths and bad punches.

Free Samples

Audio quality increases as the sampling rate goes up, so audio recorded at 48 kHz will sound a bit better than audio recorded at 44.1 kHz. The upper limit of recordable frequencies, aka the Nyquist limit, is 24 kHz for 48kHz audio and 22 kHz for 44.1kHz audio. However, if the final product is going to be at 44.1 kHz, the finished mix will need to be converted, and that can negatively affect audio quality. Having done many comparisons, I've observed that the quality of audio that's been recorded at 48 kHz and then converted to 44.1 kHz is slightly worse than that of audio that has been at 44.1 kHz all along.

Image placeholder title

FIG. 3: It''s best not to print your reverbs onto the tracks you''re sending to a mix engineer. If you really like the sound, print the reverb only onto an additional track.

If your final product is to be at 44.1 kHz, such as a CD or download, I recommend recording your basic tracks at 44.1 kHz. If the final result is to be 48 kHz, such as music for broadcast, then record your tracks at 48 kHz. If, however, the final mix is to be converted to analog for the mastering process, or if the mix engineer will be using analog gear across the mix bus, then use the highest sample rate that you and your system are comfortable using.

Inboard, Outboard

If an effect is an integral part of a particular sound — such as with an amp simulator, automated filter, tremolo or chorus — then by all means print the effect along with the sound on the track you're giving to the mixer. It's a different story with reverb, however. If there's a particular reverb you like with a particular track, print it, too, but do so on a separate track (see Fig. 3). If you combine the reverb with the dry signal, the compression that the mixer will likely use will bring up the apparent level of the reverb. When it comes to compression and EQ, it's best to print tracks without or with very little. You want to leave the mixer with as much flexibility as possible.

When you're printing stereo tracks, keep the panner (or panners, depending on your DAW) at the default position (so the sound is fully left and right). If you want your stereo track panned to one side or the other, tell the engineer that. Printing a panned track restricts the amount of stereo information available to the mixer.

Timing Is Everything

If the DAW software you are using is different than that of the mix engineer, it's best to send audio files rather than session files. Each file should start at the same place (typically bar 1, beat 1, tick 1), and it should have a timing reference on it. A short click placed one bar before the start, or a 4-click countoff, should be on every track. The click should be the same audio copied or bused onto every track.

Image placeholder title

FIG. A: The waveforms'' cycles on the left and right side of the edit point don''t match, which can cause clicks or pops.

Do not count on your DAW to start each track at exactly the same time just because they're set to do so. For example, in Digidesign Pro Tools, a bounced track can start at a slightly different time than a consolidated track, even though they were both set up to start at the same time. This is important because even slight differences in timing can play havoc with the phase relations in a mix. Having a countoff will allow the mix engineer to precisely align the tracks if there are any problems.

Use Your Ears

It's a good idea to talk to your mix engineer before you send tracks to find out how he/she wants them delivered, and then heed the advice you get. Also, check in after the tracks have been sent to make sure everything arrived intact.

Whether somebody else is mixing your project or you're doing it yourself, making sure your tracks are clean, well-recorded and well-organized will help ensure a better final product.

Steve Skinner is a producer, mixer, arranger and programmer based in New Jersey. See his record credits

Eradicate Clicks and Pops

A click or pop is usually caused by a waveform that moves instantly from one level to another (see Fig. A). If this is due to a punch or edit, a crossfade is the easiest remedy. If the crossfade doesn't work or sound good, then the solution can be to match the two files manually. Trim the ends of both files to zero-crossings (where the waveform crosses the center line when zoomed into the sample level) at spots that appear to be matching parts of the cycle, then move them together (see Fig. B). This generally will not affect the timing noticeably, but you should just move a small section of the file, leaving the rest in its original spot.

Image placeholder title

FIG. B: An edit where the cycles are better matched.

It's more difficult to find zero-crossings at the same place on both sides of a stereo file. A combination of close matches to zero-crossings combined with crossfading will generally do the trick.

If the pop or click is not at an edit or punch, then the easiest solution is to use the click-removal tools in audio-restoration software. If the software alters the sound of the file noticeably, you may need to remove the click or pop manually. Method one for doing so is to erase the pop or click. The erased area must go from one zero-crossing to another or two more pops will be created. Method two is to redraw the waveform. This works well with short clicks created by virtual instruments. Method three is to cut out the section that contains the click or pop, and re-join the sections with a crossfade. This will change the timing of the track, and should only be used in non-timing-critical spots, such as sustained vocal notes, pads or string lines.
Steve Skinner