Since we covered the hardware and software editor setups in my last installment, we’ll cover filters and techniques in this second part of the series. The main objective is to clean up your WAV file while leaving the sound quality of the file intact. Too much processing will make your sound file sound “processed”, for lack of a better word, and strip it of any life it may have.

First of all, when recording, I usually end up with a single WAV file of the entire album, or a WAV file for each side of an album. For shorter albums, I’ll just do it as one file to cut down on processing time.

For filters, I currently have the filters that ship with each of my audio applications (Sound Forge, Wavelab, Adobe Audition), plus a handful of other filters which are DirectX plugins. The best filters I own are the Waves filters–they’re a professional product. I’ve found their click and crackle filters to do the best job removing noises without leaving audible “pops” in the sound, and their EQ filters sound excellent for a digital product. (Normally you’ll hear ringing or chirping with most digital filters.) Sound Forge has other tools that do not qualify as filters…things like normalization, channel converting, resampling, bit-depth converting, etc.

First of all, I’ll load in the WAV file and take a look at it. I’ll usually spot-check it if the waveform looks odd at any point, and listen to it. This could indicate a dropout in the digital transfer. If it looks good, what I do next is check the first track on each side of the album, since those are often the noisiest. I listen to quiet passages in the first track and the others to gauge what kind of noise I’ll be dealing with. If I’m lucky, the majority of the noise will be clicks and crackles, which are the easiest to remove. If it’s steady state noise, there is not much I can do with it. If it’s too objectionable, I’ll try to locate another copy of the album and not waste any time on the WAV.

Next, I’ll go track by track and start the cleanup. First pass, I’ll look for the louder tracks with a lot of percussion, and listen. Can I hear any clicks or pops? If I hear some scattered throughout, I’ll try the Waves Click filter. What I do is highlight the track, then open the Waves X-Click filter and hit the Preview button. I’ll listen to the file using the Difference option, which lets me hear what the program is removing. If I hear ANY audio content in the result, I back off on the Threshold setting until I don’t hear any more musical clicks being removed. (This is difficult with percussion–the click filter will shave off the sharp transients of the percussion and make it sound really dull.) The Shape control lets me adjust the type of click to remove–it’s either a very sharply peaked click, or a duller one (basically, the width of the click waveform). Once I get a good removal, I’ll listen using the Audio option to hear the finished result. If the clicks are still present, I still run the filter, but then manually go in and edit out the clicks by hand using the drawing tool.

For quieter files, I use the same filter, but I’m able to use more aggressive settings since there is little or no percussion; there is also no percussion to mask some of the noise, so it will stand out more. Again, I’ll listen closely to the music and adjust, run the filter, then touch up any major ones by hand.

For songs that have very quiet sections, or a fade-out, I’ll sometimes run the Click filter on maximum settings to get rid of the “garbage”. While it might very slightly dull the sound during a fadeout (keep in mind I only use this at the tail end of it), since the level is decreasing, few if any listeners will notice its effects.

Unique to the Waves filters is a Crackle reduction filter. This gets rid of the tiny sounds that are too small for the Click filter to get rid of. A well-set Crackle filter will get rid of a ery small layer of noise and haze without noticeably affecting the music. In short, it’s fantastic! Set low, it gets rid of the tiny “zits” (my pet name) that still exist after running the Click filter And again on fade-outs and very quiet passages, you can get rid of a good portion of the noise with more aggressive settings.

For some songs, I’ll often run both of these filters more than once or twice depending on the section of the song (loud or soft). This is better than doing a single pass that either doesn’t remove the noise, or removes too much of the sound, including the music. And there is always manual removal done here as well–these filters do not get everything, and it’s not uncommon that I have to repair a file in a few places for an unidentified thump or splat of noise these filters won’t catch.

For hiss and other vinyl noise, my preferred filter is….nothing. I don’t touch it. I have yet to find a good filter to remove hiss that doesn’t also damage the music. It either removes too many of the highs and makes the music sound dull, or it sterilizes the sound too much and makes it sound processed. Or worse yet, you hear digital side-effects of the filter, giving it the effect of listening to it in a tin can. Can’t really describe it until you hear it!

For mono recordings, I sometimes use a few extra steps. If there is a nasty section I can’t repair well, the bad section is usually more in one channel than the other. What I do is copy the better channel over to the bad one, just for the duration of the bad spot. It’s seamless, and can’t be heard, and saves a lot of cleanup work. If there is a lot of groove wear, too, I can sometimes use the majority of one channel’s signal rather than use both channels full strength. Whatever works! The best part about mono is that it’s like having two separate sound files to work with, and can borrow from either channel if needed. There is another mono step I cover below.

Once the file is cleaned, I clean up the noise between tracks. For my projects, most of the time, I’ll use digital silence between the tracks. In Sound Forge, I expand the view settings so the waveform is really large in the editor, and stretch the timeline out so I can clearly see where the music starts. Rather than cut it off abruptly, I put a very tiny fade-in (we’re talking a few hundredths of a second) before the music starts so there are no pops. I then select the sound backward so it reaches the beginning of the file, then I use the Silence tool to kill all the sound. From there I go to each band between the music tracks and clean those up. I fade out each ending. If it’s an abrupt ending, I fade out the sound within the echo, usually the last half of it so the faint music and vinyl noise both dissolve into nothing at the same time. It’s very true to the original fadeout. If the song itself fades out, I do it similarly. I usually locate where the faded out track’s sound ends absolutely, then set my fade-out ending point just past that. Then I go backwards a bit and find the approximate halfway point and start my fadeout there. In most cases, you would be hard pressed to notice any difference between my edited, cleaned-up fadeouts and the originals. To finish up the cleanup, I trim all but about 0.2 seconds off the beginning of the file, and maybe leave one-half to one second of silence at the end.

If the track needs any parametric EQ, I’ll do it here with the Waves Q-series Paragraphic EQs. I rarely need it. If there is a nasty peak to get rid of, I’ll pick a very narrow Q (band) of EQ, set it to boost by about 12dB, then move the slider up and down the frequency range until I pinpoint the central spot with the response peak. From there, I can then cut the peak down and adjust the Q so it gives the range a gentle nudge downward. And a good rule of thumb is to pick your EQ setting, listen to it, like it, approve it…then cut your adjustment in half. I do not like to tilt the low or high end up for the most part, unless a recording is really bad. EQ, like any filter, is best used sparingly.

Is the original recording mono or stereo? If it’s mono, I next do a channel conversion–basically, summing both channels together and creating two identical channels. (Each new channel contains 50% left and 50% right information…but that ratio can be adjusted for problems with the recording, as noted above.) Doing this removes a LOT of vinyl noise, since a lot of it is phase-related. Plus, in the stereo spread, the noise is masked in the center by all the center-channel information. Mono recordings often clean up much better than stereo due to this simple channel conversion!

The final step is volume adjustment. Rule number one: stay away from compression! For a needle drop, I will only use the normalization tool. Normalization increases the volume level until the highest peak is just at the 0dB digital mark. This makes full use of digital’s dynamic range. You cannot go over 0dB in digital without extreme distortion. If you have a track going into a compilation that just isn’t loud enough, you can use a compressor filter…but only use it as a last resort. The better compressors don’t leave much of an effect on the music. My favorite is the Waves L2 Ultramaximizer. On its lower settings, you hardly know it’s being used.

If everything is OK, I’ll save the file one last time (saving a copy, of course, not the original), then do a downsampling and downconversion to 44.1/16. I’ll save that as a separate file.

From there, we go to CD Architect and drop the files in place. For track markers, I usually put them about 0.2 seconds before the music starts. I always test the track markers a few times before burning the final disc.

And that’s about it! We burn the CD, and we play it back! If all has gone well, it should sound pretty good. Put it away for a few weeks and listen to it again when you’re not sick of hearing the music repeated over and over…and it should be a nice reward for all of the hard work!