This is a list of the factors related to recording and encoding files I think may make a difference in getting iTunes Match to, well, match recordings made from analog source material, specifically vinyl, though some people have thrown cassette recordings at the matching process and had some success. It's hardly comprehensive and certainly not all my own. Roebeet in particular has provided quite a bit of detail on his process and success in getting things to match. I just wanted to summarize my thinking.
First off, nothing I write here is meant to criticise anyone's way of doing things and it's certainly subject to change. My success is far from 100%, though it's improving. My analog equipment isn't anything to write home about either, though as I can, I improve it too.
Comments and observations from Roebeet have certainly expanded my mind on the subject in the last few days, and the changes I've made in my process thanks to his input does seem to have increased my matches. It's also sped up the process of getting vinyl into iTunes, match or no, and I'm thankful for that.
As has been said by many, changes Apple may make in the future might result in our getting better matches. However, it is also possible that when it comes to files from analog source material that changes on Apple's end might result in fewer matches. I can't do much about what Apple does (those who regularly talk to people above L1, thanks though), so I choose to work on my process to see if I can improve my success rate. I do it in part because, hey, it's kind of a fun challenge for me.
First, try to put aside any successes you've had getting other matching services to match your songs. Apple's matching to what is in iTunes, which means at best it is using a subset of the larger Gracenote database other matching services use. Further, when it comes to matching vinyl, what appears to be the "same song" for sale on iTunes, in reality is at best similar (more on that later). Finally, we don't know what Apple's agreement with the RIAA is, but chances are the deal that lets us download better copies of thousands of songs for chump change forced them to set the bar pretty high. I'm of the mind that they were right to set the bar high, then work to improve matching, as to do it the other way would likely set off a firestorm.
OK so here goes . . .
• Be certain your turntable is operating at the right speed and that wow and flutter aren't coming into play in a major way. You may assume that your TT is fine, but if you have never put a strobe disk on it, just taking it for granted because of a speed lock light or something is a little presumptuous I think. You can download a strobe from VinylEngine among other places. The thinking here is that deviations in speed that you may not hear can compress or expand the waveform enough to preclude a match. I now check before each recording session and the slight drift I need to correct for (I'm using a Dual 1219, which is an idler wheel with a speed control knob) is slight, but it is there.
• Work in lossless formats throughout. Using a lossy format for the final output file, the one you put into your library and submit to iTunes Match is fine and may for you be preferable. I use lossy at the end because I don't want to choke up my library (in particular those uploaded to the cloud and back down to iPhone). The thinking here is that you lose some data any time you convert/export using a lossy format. Some people move in and out of programs in order to do noise removal, so if done in lossy, each import/export could result in additional data loss. You may not hear it, but iTunes Match may need that extra data to do its thing.
However, when it comes to your final output, beware that not all AAC files are the same. My own experience is that some AAC files created with encoders outside of iTunes will get through Apple's initial screening process, may even result in matches, but if they are uploaded, they will not play. The encoders I've had trouble with have been tied to Audacity. I now use 16-bit WAV until the end which has the added benefit of speeding up my work. The number crunching done to create an AAC is a real time waster. You may have a lot more horsepower than I, but individual tracks exported from Audacity in 16-bit WAV get done in seconds, where AAC files take me 2-4 minutes. It's one place where speeding things up actually produces better results.
• Beware of the potential for clipping to be introduced at points where you may never see it. As Roebeet has pointed out, people using programs like Audacity have learned to keep initial recording levels low (-6 db is often mentioned), then they use the Amplify function to take things up to 0 db. Problem with this is that you've left yourself no headroom. The file may be clipped when converted during an export. Robeet's looked at a lot of maxed out files in Audacity and has seen plenty of evidence that this happens. His solution is to set Amplify at a -0.5 db, something I've already added to my process.
• Set your label tags as close to the beginning of each track as you can. I used to be a bit sloppy, just made sure I didn't clip the track. I don't know if there's an industry standard for this, nor whether such a standard is applied to everything in the iTunes store, but it stands to reason if the matching process involves locating the true beginning of the song and from there matching it to Apple's baseline copy, best not to give it anything but true music to look at. While Audacity has a "find silence" feature that can put blank tags before a lot of tracks, it's caused me enough grief that I set all my tags manually. I can do it faster anyway.
• Give some thought to how you're doing your analog to digital conversion. If you have a high-end sound card or breakout box, you may be fine. If you feed a preamp out directly into a Macbook and use the laptop's sound card, you may have a problem (that's me, by the way).
The biggest issue I'm aware of is a DC bias offset introduced by the sound card. Audacity has a filter to correct this. Before running it, areas of silence that are positive on one track are negative on the other. After running it, the areas of silence look a lot more like silence. It affects the waveform significantly. I should say I've never put anything to the match process without running DC Offset so I don't really have any evidence to support that it matters.
• Some cleanup is going to help, some is going to hurt. Of course, proper cleanup that actually involves cleaning the record before playing it is always a good idea! If you've bee working with vinyl for some time, this is a no-brainer. If you are new however, do yourself a favor and read up on proper cleaning techniques, as some not so proper methods may damage the record or leave a residue you don't need the hearing of a hunting dog to pick up on.
When it comes to digital cleanup, err on the side of caution, but know if done right, it may get you a match. It seems most people who start out in Audacity use its Click Removal filter, but after awhile their ears tell them a lot more than just clicks and pops are being removed and move on to some other program. I use ClickRepair and run it in a semi-auto mode that does a good job of helping me keep from removing transients. I'd put removing transients into the same category as clipping the signal - you're making significant deviations from the original that may easily kill a match.
I have some experience that suggest noise can kill a match, particularly on songs with quiet passages. For example,Track 1 on Zeppelin's In Through The Out Door, which starts with a long, long, long fade up, did not match for me initially. Listening to it, I was surprised at how much noisy my copy was. Still,it did match after I went back and reprocessed using ClickRepair.
• Resist the temptation to replace the silence between tracks with true silence. Mess with the dead air and you risk altering the beginning and end of the music. Some of my own experiences suggest fadeouts may hinder matching. For instance, I've got two albums by the same artist, recorded the same day using the same workflow. Every song on both albums, save one, matched. The last song on the B side, which is the only one I normally use a fade out on. If you just can't stand all the noise between your tracks, consider that a sign that the tracks are probably full of enough noise to preclude a match anyway and take steps to correct the problem.
• Do all you can to assure your metadata is as complete and correct as you possibly can make it. While it's clear that the matching process does not preclude even unnamed tracks from getting matched (I'm one of many who have had several unnamed tracks make it through), the first step in the matching process involves gathering information from your library. While I've seen nothing that confirms what Apple is doing with the information it gathers, my assumption is that at a minimum, an attempt is made to first identify possible matches based on song/artist/album title, then to actually compare the songs to see if they match. If this is true, then at the very least, the more accurate information Apple's got to work from, the faster the process will be.
If you haven't run a program like TuneUp, you might want to do so. I haven't been nearly as diligent as many are about keeping my tags clean, so the process of getting things through iTunes Match has been going on at the same time I've been cleaning up my library - not good.
• Be prepared for iTunes to throw you a lot of crazy matches. Some curve balls have been well documented: Explicit versions being matched with clean ones, album-length songs being matched with top 40 shorty versions, monos matched with stereos, etc. This is happening regardless of the media the source material came from. I think we can all agree this is a pretty big negative that Apple needs to correct.
On the plus side though, if you want access to a better copy of a song than what you have, if Apple has that song on a compliation, or anywhere else for that matter, it may match it. In doing so, it may show up as being on an album other than the one you've recorded though. A prime example for me is a recording I made from a Verve compliation of Count Basie songs from 1955-56. iTunes doesn't have it. In fact, Tuneup didn't see it either, but it found a number of the tracks on another Basie compliation (Though some might be renditions done on a different date - I haven't checked yet). I decided to go with Tuneup's tags. All of those tracks ended up being matched.
• Finally, don't expect miracles.
If you are working from vinyl, nothing that Apple has on its servers is the same as what you recorded. The music industry didn't make all those Beatle's digital files by dropping a needle on a record and even if it had, the chances are the equipment they used would be light years ahead of what most of us are using (though for some recordings, a good argument can be made that they still didn't do a very good job).
Be thankful that Apple's matching system needs to be loose enough to match files that aren't identical to the baseline file. After all, while vinyl is unique, the same can be said for all those low bitrate songs people are looking to upgrade, it's the need to accommodate sub-par all-digital rips that's allowing any vinyl to get through at all. I've certainly not seen any language from Apple suggesting the service is supposed to let you get high birate copies of songs you recorded from vinyl, cassette, 8-track . . .
Personally, while I do want more matches, I'm pretty happy with my results so far. That I can access my entire music library on any of my devices is a big deal for me. Best of all, I no longer have to hit myself in the head when my music suddenly stops because I forgot and closed the laptop!
