How I Made My Recordings

by Mark Woodford July 28, 2000

last update August 8, 2000

A few people have asked me how I made these recordings, so I decided to write it down....

First let me say that I don't consider myself an expert. I'm just a software engineer with a love of singing, playing guitar, and playing with my computer. I do this entire process on my PC so I get to do all three. How Fun!

I'd like to thank all the folks on rec.audio.pro who take the time to help those of us who are very new to this process. I'm just a wee lurker in that newsgroup, but I learned a lot just by reading your ramblings. I'd also like to thank the deja news power search service for helping me sift through rec.audio.pro.

My recipe:

Make a MIDI rhythm accompaniment (sometimes with bass) for the whole song including intro and ending.

I usually use Jammer Pro by SoundTrek to generate the first draft, then use Cakewalk to clean it up, remove any parts I don't like, add a high hat lead-in, etc. Jammer Pro is similar to Band-In-A-Box.

Render (record) the midi file as a .wav file

I used to use my Tropez+ for this, but lately I've been using SeerMusic's software synth to render the midi file directly to wav file. Great sounds, Very clean, no noise.

If you use a combined midi/wave multi-tracker like Cakewalk Pro Audio you won't need to do this, but I haven't had much luck syncing midi and audio, so I just turn the midi into audio and stick to syncing audio to audio.

Load the .wav file into a PC based multi-track recorder.

I used to use SAW+ (Software Audio Workshop Plus) by Innovative Quality Software as my multitrack recorder.

Lately I switched to a great shareware program called n-Track Studio by FaSoft. It works great, is much easier to use, and only costs 40 bucks. Great Deal! Highly recommended!

Make sure you have a very fast hard drive that is not fragmented and has lots of space. CD quality 44khz stereo audio takes about 1 megabyte per minute. A 4 minute song with 6 stereo tracks needs 24 megs for the tracks plus 4 megs for the mixdown. With retakes you get to 100 megs very fast.

Record a bass track and 1 or 2 rhythm guitar tracks while playing along with the midi wav track.

SAW+ or n-Tracks is my multi-track recording software.

You need a soundcard that has high quality A/D converters and can tightly synchronize audio while playing and recording at the same time. Cards that don't sync properly may make your recording life miserable.

I use GadgetLabs Wave/4 as a 4 channel (2 stereo channel) simultaneous play/record sound card.

I've recorded guitar and bass lots of different ways and they all seem to work pretty well. Recently I bought a Roland VG8 guitar synth that makes guitar recording very easy. It makes almost any guitar sound, and what you hear from your monitor or head phones is what you record. Highly recommended.

I've also used and liked SM57 in front of loud guitar amp, SansAmp Classic, for recording bass and distored guitar, Piezo acoustic on one channel and mic'ed acoustic on the other channel.

I use Lexicon MXP100 for a little reverb. Sounds very nice. The tails aren't too grainy.

Use a sound editor to compress, normalize, and clean up any noise on all the tracks recorded so far.

Using Sound Forge 4.0 by Sonic Foundry for this I usually do the following:

Mute the beginning and end of the track.
Clean up any other unwanted noise in the file.
Normalize the whole file, making the loudest sample 0db
Gently (2 or 3 to 1) compress as needed to get rid of any very loud spikes in the file.
Normalize to 0db again after the compression.
Repeat 4 and 5 as needed until the track is suitably compressed.

Adjust the levels in the multi-track recording program so the "band" sounds pretty well balanced, not finished and not micro adjusted, just tolerable.

Back in SAW+ or n-Track, adjust the faders so the band sounds reasonable enough to sing and play along with.

Record a vocal track while singing along with my multi-track recorder "band".

I've been using a RODE NT-1 for "pretty" vocals and an SM57 for louder, more rugged vocals.

The mic is connected to a JoeMeek VC-3 for pre-amp and mild compression. I don't use the enhancer function.

I don't use any effects on the recorded vocal. Reverb is added later.

I don't have a special room, so I just make sure the furnace isn't running and the kids are either gone or sleeping. Lately I've been using cordless headphones. This is nice when you're running back and forth from the computer to the mic.

Record one or two lead guitar tracks while playing along with my "band" and vocal tracks.

This process is the same as with the rhythm guitar.

Use a sound editor to compress, normalize, and clean up any noise on the Vocal and Lead Guitar tracks.

This process is the same as with the rhythm tracks, i.e. clean up, normalize, compress, normalize.

Record a reverb track for the vocal only, in the multi-track recorder and adjust the general reverb level to sound balanced.

I set the Vocal track(s) on Solo and play them into the Lexicon MPX100 with the Plate Vocal setting, recording directly back to the sound card.

Adjust the whole mix.

At this point I sit in front of SAW+ or n-Track for a long time tweaking the volume of every track throughout the song. Both programs "remember" and display the fader setting for each track at each time in the mix. I think this is what they mean by mix automation. I'm in awe thinking that some people can do this in real time during the mixdown.

Some of my favorite techniques are:

Make sure stupid parts and otherwise undesirable parts are turned off. (duh!)
Make sure the vocal is the loudest track. (usually)
Lower the volume 3-6 db on the guitar and keyboard parts every time a vocal comes in, and raise it back up when the vocal finishes. Kind of like compressor ducking. This allows the band to stay seeming pretty loud, but the vocals still come through clearly. You can do this on bass and drums too but I think that sounds funny. Ducking the guitar, piano, and other midrange tracks sounds best to me.
Lower the vocal reverb level on "intimate" parts.
Raise the vocal reverb level on "BIG" parts.
In a part that seems stale, slightly change panning and level on a few tracks.

Mixdown from the multi-track recorder to one stereo wav file.

SAW+ and n-Tracks render the mix directly to disk as a wav file so this is very easy.

Compress and normalize and clean up the beginning and end of the file.

Using Sound Forge I do the same technique described above to get rid of any unwanted large peaks and maximize the loudest sample to 0db. Compress, normalize, listen, compress, normalize, listen, etc... I usually compress a lot on the big loud stuff and not so much on the quieter delicate material or songs with more space.

Burn a CD.

I use Sony Spress 2x/8x IDE with Adaptec EZ-CD Pro. This old work horse has served me well.

Make an mp3 file and put it on my web site.

I use the very simple BladeEnc for DOS encoder.