Seamless Loops and Splicing

attention: musichack is moving! that’s right, we’re moving out of the wordpress apartment and in to our own server…sorta (ok, so I run other domains..but hey, i pay for the damn thing)…but I’ve installed wordpress-mu and will be moving the entire ordeal there. the address is also going to change…rather than being a subdomain…we now have a domain. http://www.musichack.org will be where you’ll find the new musichack..with all the old content and tons of new stuff. check the main page for the final notification about the move, also, the notice about “Gold Wave Editor” has been removed since it appears they have changed names…but still be on the lookout for the editor with the space in the name…that looks nothing like my screen shots.

The first article in my Goldwave series involves the process of both locating and creating loopable samples..and what I call “seamless” splices..audio splices that appear to be completely “transparent”, as in, you can’t noticably tell there’s a splice there. This was one of the first things I did in audio-manupulation that made me go wow and “if I can splice something this smooth, what else I can do?”. In fact, the other day I was faced with this exact same task and, not having done it in a while, I decided to document the process and share it with you.

The scenario is this; I was going through some transfers from some old reel-to-reel tapes that were given to me when I recieved my RTR deck and decided to at least make a backup copy of the analog tapes for the original owner. Naturally, while going through these tapes I found some interesting songs that are thought to be “practically lost”, and a couple I thought were purely interesting.

On the tape there’s a jazz-like cover of The Theme from Shaft, and it was a rather decent recording. The problem is (as with a lot of amateur recorded tapes), was that in the intro there was a bit of “mis-recording”, a .7 second portion of audio recorded over the track. When dealing with tape restoration (which I will cover some aspects of in later articles) this can sometimes be what would be considered “death” for a track, however, in some cases the original portion of audio that was recorded can somehow be “replaced” with elements from later on in the track..this is common for a LOT of songs. Sometimes it’s not so straight forward..and I’m not going to be covering how to repair EVERY complex portion of audio you need to..I’m only here to show you the basics…once you have the basic method down you can expand upon that rather easily to fix more complex things. While having the basic skills will help, you also need to listen to the audio and, at least in your head, try to assemble WHAT it should sound like, and if you can wrap your head around audio and “feel” it, it’ll become easier. This will happen either after years of playing around (in which you’ll gain an understanding), or sometimes your brain just clicks on it (i call that “natural talent”).

The example audio files for this tutorial are contained in edit-splice.zip and contain two ogg files; the original (before) and the fix (after). You can use the original file to try this project yourself.

As you can see, the portion of audio that doesn’t belong tends to stick out, it’s a much louder blob in than the rest of the intro. If you listen to the into you’ll find it’s a rymthic pattern of cymballs that repeat and then an additional set of cymballs join the mix. This is a rather standard method of starting a track (the more you listen to music and pay attention to mixing/mastering, you’ll hear it) and actually, in this example, is PERFECT for a beginning splicing project. The portion of audio missing comes entirely before the second set of cymballs come into the mix. Listening to the first portion of the intro, it’s obvious that it repeats, and there’s enough of the audio before the recorded portion to re-create the lost portion..so let’s get to it.

From zooming in vertically we can clearly see in the waveform the repeating pattern, but you should NEVER go solely on waveform alone, you should listen to the audio to make sure that’s in fact here a “clean” loop will occur..often drums will loop perfectly but there’s other audio elements on top that won’t. In our case, there’s no drums, and can use the waveforms to help us isolate a loop.

As you can see from this screenshot we’ve zoomed in even more around the selection of audio we’re going to be using for our loop, but we haven’t yet finalized the positioning. Now, the basic idea is to go from the beginning of the attack to the begging of the next major attack (the inital cymball crashes)…which on the waveform are the two large waves that seem to have a flat side. Zoom in to the side of one and look for where the wave starts changing for the attack.

It’s always best if you can pick a position at the “zero-crossing”, which is where the waveform crosses 0 amplitude; the reasoning for this is if you cut a splice a section of audio and the sample before the splice and first sample of the splice have MAJOR differences in amplitude, it will create a click that you’ll either have to figure out how to manually get rid of or try your cut again, however, in most cases, you can do a bit after splicing to fix this…but more on that later. Now that we’ve got our beginning position set, look around to make sure you’ll recgonize about the same spot when you pick your end-point. At the current zoom-level, I chose to pick right before the waveform modulation becomes recgonizble (there’s additional movement, but, at that zoom level it looks mostly like 0, and in reality, it’s generally NOT that critical). Zooming in to about the end point of selection I’ll place my end marker at a similar (because, it’s not going to be EXACT) formation…which I’ve done in the screenshot.

Now you have your selection made, and in our file, it’s .446 seconds long, but does it loop? There’s an easy way to tell. If you haven’t reprogrammed your play buttons, right now, right click on the yellow button, put it on selection and make sure loop is checked. This will set the yellow play button to loop your selection. If you’ve done it right..it should sound loop seamlessly without any popping or clicking or noticable looping position (the position where the selection ends and it starts over). Sometimes previewing your loops this way will ensure that when you get ready to paste them, they’ll loop seamlessly.

Now that you’ve got your loop selection, copy it to the clipboard..and just for safety sake (so you don’t lose it), go ahead and click Paste New (P.New) so you’ll have it in it’s own window.

Now we need to work on removing the bad audio and creating the new selection. Now, the method I use for doing this is my own, and I’ve found that by doing it this way..it’s slightly more complicated since you’re dealing with more windows, but you can tweak the final selection before you paste it in.

Looking at the screenshot we see there’s about the amount of one loop, the record-over, and another loop starting not long after it. Rather than dealing directly with the damaged audio (and having to deal with half-loops and such), it’s easier to keep cutting and splicing on the “beat-point”, so that’s what we’ll do, we’ll do that by selecting the last cymball crash before the glitch, and ending at a similar point after the glitch, and again, this decision will play a lot on your interpertation of audio and how your brain works. As you can see, I’ve already selected the portion of audio we’re going to discard and replace. Now it’s time to create the replacement. As sated before, our loop is .446 seconds long, and the portion of audio we’re replacing is 1.374..so we’re gonna have to paste the loop a few times. Go back to that loop you pasted into the new window and, assuming you haven’t copied anything else to clipboard, click paste. This will paste the loop portion in the same window again..and brings us up to .892, which isn’t enough. 3 pastes will give us 1.339 and 4 will give us 1.784. Neither of these is EXACTLY the same as 1.374…but 1.339 is close enough. so, if you haven’t clicked Paste already, click it twice so you have a file that contains 3 of the looped portions looped together. Make sure the entire window is selected and click copy. This puts our 1.339s loop into the clipboard, overwriting the original .446s segment we created it from.

Now, hopefully you haven’t moved the start/finish makers on your original file, becuase we’re ready for the magic. Go to the original file and hit delete (either click Del on the toolbar or, my favorite, the delete button on the keyboard). The selection will get deleted and you’ll find that actually, the file looks like it’ll play seamlessly that way, and it will..but it won’t be the right length/timing. Don’t move the start/finish markers from thier position, right now they’re at the exact point we want to insert the loop.

Click Paste (Not P.New or Mix, but Paste). That’s it! It will insert the loop at the position the bad audio was removed.

Now you can zoom out, select the entire intro and listen to what should sound like a completely restored intro. If you can barely tell where you did your cuts, then you’ve obviously done a great job..if you can tell, well, you’re probably a few samples off and it might take you a while to get the hang of it.

*NOTE* I forgot to mention this the first go around, but there is some bleedthrough noise on the tape…where the recording on the adjacent tracks is bleeding through to the tracks being played back….if you intently focus on that, you’ll be able to hear the studder and know where the splice is…however, that’s a defect we have to live with..it exists in my finished sample as well.

This is just a basic skill…it shows you that it’s possible to make seamless splices and hopefully gave you enough information on how to detect them, but it doesn’t have to end there…you don’t have to deal with loops persay…you can actually edit songs down this way. About a year or so ago, I posted an iPod commercial to YouTube…it was The Fratellis’ Flathead iPod commercial. (http://youtube.com/watch?v=PKxGfLo7Cqo) One thing I did (in an attempt to improve upon the original tv audio), was to recreate the commercial edit entirely from audio-cd (or in this case, FLAC) source. Again, it was a lot of intent listening to the tv edit and CD version to find out where they pulled elements from and how they were assembled. I don’t remember exacty how the intro was edited, I beleive that was a simple cut/splice..the little cymball-ding was actually pulled from an entirely DIFFERENT section of the song and “mixed” in at the right spot. The ending is SOMEHWAT obvious. When the guys start going “ba-da-da” (or something similar) when the iPod + iTunes card pops up, the first part of that is from the previous audio, the ending…where the additional backing vocals seem to kind of come out of no where..that was spliced practically mid-word from the ending. I also at one point took “The Hives – Hate To Say I Told You So” down to about 45 seconds without it being obvious it was a cut; I had to extend Louie Armstrong’s “It’s A Wonderful World” for play at a wedding reception by splicing the first half of the song back over the second half and reassembling elements to make it wound natural…once you’re able to wrap your head around this simple concept and start hearing how the things should sound..then start to identify where to make the cuts, splices and such, you can do some really cool things.

This is what I consider the first basic building block of audio manipuation..you don’t need to do it often, but as I showed here..sometimes it can come in handy.

This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 Unported License. GoldWave® is a registered trademark of GoldWave Inc. Screenshots courtsey Goldwave, Inc. Used with permission. Created 09 Jun 2008 by Jay Moore (dewdude@g*oogle*mail.com)