VividCon 2006: Audio Editing Panel notes and examples
My own notes for this panel resemble my notes for teaching, which is to say that they are extensive and yet idiosyncratic to the point of being incomprehensible to anyone who isn't, well, me. I have attempted to arrange and elaborate on them in a way that might actually be readable, and possibly even useful.
I've been thinking a lot about audio editing since VividCon 2004, when several different people, talking about different vids of mine, mentioned song choice as something they thought I did well. The compliments pleased me very much, especially because song choice was in fact one of the elements I was happiest with in each of the vids in question. But they also made me chuckle a bit, because I didn't just choose those songs; I also, to some extent, built them - by which I mean that I manipulated them in various ways, from cutting whole verses to trimming phrases.
Put simply: while some songs may in fact be perfect for vidding purposes, many need a little help.
So, as I said at the beginning of the session, this was not a panel about whether to audio edit; my assumption was, and is, that audio editing is always a good thing to consider, and often a good thing to do.
NOTES ON EXAMPLES
As I said at the panel, I've drawn all examples from my own vids not because I'm the most egotistical vidder ever (...although that may well be true) but because the whole point of audio editing is that listeners shouldn't know that anything's been altered (unless, of course, they know the original song very well). So these are the audio edits I know best (and can easily export for examples). For a very recent example of extremely successful audio editing, check out sdwolfpup's "Fix You", a 2.46 vid based on a 4.55 song.
REASONS TO EDIT
I know several vidders brought up other excellent reasons for editing, but I didn't write them down and my memory's shot; if anyone remembers what they were, comment and I'll add them!
TOOLS FOR EDITING
Any decent video editing software (Premiere, FinalCut, etc.) will also allow for quite a lot of audio editing: snipping out chunks of the song, putting bits on different tracks and crossfading them to blend the sound if necessary. Until very recently, I did all my audio editing in Premiere - including the rather nitpicky cutting of "Thistledown Tears" (see below). As killabeez observed during the session, editing audio within the vid project has the advantage of allowing us to change our minds while vidding, to tinker with the song the same way we tinker with our clips.
There's also the option of using audio editing software; I've used the evaluation version of Goldwave, for example, and have recently started using Adobe Audition (which I love, so if you're looking for a recommendation, there you go). Several panel attendees mentioned other audio editing software they've used successfully, so ask around if you're interested. This type of software has the advantage of offering more effects (and particularly more effect presets) than video editing software. For most of us, I think any basic editor (preferably but not necessarily multitrack) will work just fine, although I do recommend making sure the software has a center channel extractor (sometimes called a "karaoke" setting) - more on this in the "Out Here" examples. Unlike Killa, I prefer to have the audio edits completed before I ever start vidding, so editing the .wav file in advance and importing that edited version into Premiere works just fine for me.
TYPES OF AUDIO EDIT
NB: These aren't technical terms at all; I'm just trying to keep terms consistent within this set of explanations.
MAKING SEAMLESS EDITS
Making seamless edits requires two things:
The musical pattern part is relatively simple; we have to figure out how many measures to cut to preserve whatever musical or rhythmic line the song has established, but once we know that, it's pretty much a matter of a couple of razor cuts. (See the section on technical tricks for more about the exact how and where of the razor cuts.)
The sound part can be fairly simple or a huge pain depending on the type of music being used. Techno, for example, is extremely forgiving of audio edits; straight-ahead rock songs with a strong repetitive drumbeat and bass guitar line are generally pretty easy to edit as well. Guitar solos, however, can be a big pain in the ass, and anything with piano, strings, or noodly acoustic guitar is likely to be a bit tricky (though not impossible!). For the latter types of songs, attempts to trim and/or rearrange will probably benefit from having a multi-track editor with special effects capabilities.
TRIMMING EXAMPLE: Window of Opportunity
This song is originally 4.02; I cut it down to 3.01. That's a quarter of the song gone. Trust me when I say you don't miss it.
If you know the vid and not the original song, you know the first verse as:
I'm on vacation from the measurement of quality
backing away from sleepy hollow front lawn of tranquility
just singing in my head and wreaking havoc with authority
I've fallen out the window of opportunity
But check out the first 40 seconds of the original [right-click to save].
If you know Wonderfalls at all, you know that the lyrics of this original first verse are actually kind of terrific for Jaye, but they tell such a specific story that illustrating them with the available source was going to be challenging if not impossible. And since I knew I wanted to get the song shorter anyway, this verse got the axe straightaway.
If you count off the measures before the lyrics kick in (and pay attention to the bass line), you'll find that the song starts with a pair of four-measure patterns, which adds up to a nicely balanced eight-measure pattern. Eight measures is (as yhlee pointed out during the panel) a pretty standard musical phrase in Western music, although patterns may also come in two- or four- or sixteen-measure chunks (and, of course, in the absence of written notation, the number of measures depends on how we count, which may vary among listeners).
So one editing option would be to preserve the original eight-measure intro, and then just splice in the second verse instead of continuing with the first verse, which would sound like this:
Sample Window of Opportunity edit [right-click to save]
The problem with this strategy, to my mind, is that the two four-measure patterns of the intro sound almost exactly like each other, whereas the little four-measure musical interlude between first and second verses, though it maintains exactly the same bass and keyboard stuff, adds in drums and a bit of guitar noodling around the edges.
So, for my vid edit, I built a new intro: I used the first four measures of the original song, and then spliced in the four measures right before the second verse begins:
Actual Window of Opportunity edit [right-click to save]
Either of these edits is very simple to make: two snips, cut out the stuff in the middle, scoot the two bits up against each other, and hey presto, off and running.
TRIMMING EXAMPLE: Thistledown Tears
I didn't use this example at VividCon because we didn't have time; I'm including it here as an example of smaller-scale trimming.
"Thistledown Tears" is originally 5.13; I cut it to 3.50 (and in retrospect should probably have cut more; ah well, lessons learned). Unlike "Window," where I got rid of most of the length in a couple of large cuts, I trimmed "Thistledown" by cutting out eleven smaller chunks, some of which are only two or three measures long.
One of the things I love about "Thistledown" as a song is that it's very slightly and deliberately off-kilter in terms of its musical patterns. Where we would normally expect four- or eight-measure instrumental interludes between verses, we sometimes get five or seven measures instead. The song is also characterized by repetition of the line "night is quickly passing." Both of those elements of the song, though fun in and of themselves, are problematic in a vidding context. (Think River: "This song is problematic.") I felt that the unbalanced instrumental bits might be more distracting than interesting, and I was quite sure that, as much as I loved the repeated line and all the shades of longing and hope and weariness that Jeff brings to it, I wasn't going to be able to do something interesting with it as many times as the song included it. (Plus, he frequently eliminates it himself when he plays it live; some of the live versions end up remarkably like the edited version I made for the vid.)
Anyway... back to odd-numbered patterns and inconvenient repetitions. This snippet from the original song [right-click to save] shows both these problem elements. This snippet from the vid version shows what I did with it: cut the first iteration of "night is quickly passing," kept the first two measures of the five-measure interlude, and went right into the "hymn songs" lyric.
This bit is what I was thinking of when I said to morgandawn, during the panel, that edits don't necessarily have to maintain eight-measure patterns; in this case, we get a balanced sequence of four measures once the "extra" measure is cut out, and then I cut that pattern in half because I felt that for this particular guitar line even two measures can stand on their own as a complete musical idea: one measure of melody rising, the second falling. I did preserve the full four-measure sequence at the very beginning of the song, because it feels more complete, and I kept it again with the sequence right after the musical climax of the song, where I wanted some recovery time and also more of a bridge effect. But for interludes between verses, I decided that two measures would do - and I stuck to this throughout the song, trimming every interlude down to two measures (except at the bridge).
Because "Thistledown" is all about the acoustic guitar, the sounds are not quite as sharp and definite as they would be in more electric or electronic songs, which meant that a lot of the cuts (though not all of them) sounded better when I used tiny crossfades - and by tiny I mean "less than a second." Audition provides several different kinds of crossfade, including sinusoidal (sine wave or S-shaped) and logarithmic (a steep curve), and the logarithmic in particular can come in handy, but most of the time I find that a simple linear crossfade sounds just right. Here's a screencap of one edit and the corresponding musical snippet.
REARRANGING EXAMPLE: Out Here
It's actually rather difficult, I think, to compare versions of a song one doesn't necessarily know, especially on a single listen, so I offer instead a visual representation of the edits I made to "Out Here": a side-by-side comparison of the lyrics. Verses I cut are grayed down; the section I moved is highlighted.
Decisions about what lyrics to cut from a song will almost always be based partly on content; in this case, I cut the two verses I did largely because when I started editing I already knew that the song was going to have to reverse the meaning of the lyrics. "Out Here" is, as Peter once said, about going crazy, cutting off your hair, and moving to rural Massachusetts. For the purposes of my vid, it had to be about getting kicked out of Canada and exiled to the metaphorical wilderness of Chicago, which meant that references to actual rural locations ("got your rocks, got your rivers, got your trees") were good things to get rid of. I mean, I might have been able to make them work, but I didn't need to; the vid idea certainly didn't require it.
But decisions can't just be about content; there's also the question of structure. Each of this song's verses is in two halves that do just slightly different things musically; cut one half, and the verse sounds wrong, unbalanced. (There's also a signal in the lyrics: in the original, each of the first three verses ends with "miles away.") So in order to cut the second half of the first verse, I needed to also cut the first half of the second verse and simply smoosh the remaining halves together into a new combination. No problem.
Except I actually needed the first half of the second verse, especially the lines about "it is my privilege to lie awake at night and think of what I've lost," because if that's not Fraser in a nutshell I really don't know what is. The obvious solution was to substitute this section for the "out here it's not as bad as I tell it" section; it even worked structurally, since they're both first halves.
But when I simply spliced that section in right after the end of the guitar solo, I got something that didn't work [right-click to save]. The musical pattern's right, but the sound is all wrong: the sustained guitar note simply stops, and the sound, which has been getting fuller and louder all through the solo, drops way back.
I realized that what I needed to do was keep the first line after the guitar solo and layer it on top of the verse I was putting in its place. Except I also knew that wouldn't work, because of the lyrics. Without any adjustments, what I got was a muddy mess [right-click to save]. The guitar line does continue the way it should, but of course so does that second layer of vocals.
Clearly what I needed to do was strip out the vocals, leaving just the electric guitar and enough of the acoustic guitar to amplify the sound a bit. I poked around a bit in the tools, but couldn't find anything that was obviously what I needed; so I went to the help files, searched for "remove vocals," and discovered the Center Channel Extractor filter (sometimes called "karaoke" for obvious reasons). At the panel, absolut3destiny explained how this filter actually works; all I know is that when I applied it to my little post-guitar-solo snippet, using the male vocalist setting, I got this useful bit of sound [right-click to save].
This works. The electric guitar is coming through clearly, as are the lowest notes of the acoustic guitar; the vocals aren't entirely gone, but they're sufficiently echo-y that they aren't terribly noticeable when combined with the spliced-in verse [right-click to save]. I didn't even have to use any fades, although I would have messed around with that if the sound still seemed to need blending.
One thing I want to emphasize about this rearranging process: nothing about it was difficult. It took a while, sure, both to make the decisions about what needed doing and to figure out which of the effects at my disposal would do what I wanted; and of course there was some fiddling with settings and clip boundaries to get things to come out exactly right. But, let's face it, anybody who's not willing to take on a little bit of fiddling has probably not taken up vidding.
DISCUSSION: Tre Sorelle
In this part of the panel, I wanted to provide a hypothetical vidsong and give folks a chance to think about what they might cut if they were actually vidding it. I actually have vidded this song, and I made a particular set of choices about how to edit it, but they're not the only possible choices; there are lots of potentially right answers to the question of what to cut. There is also, however, a WRONG answer, which would be "don't cut anything". This song is techno - Italian folk-techno, granted, but still techno: it features bagpipes and violin, and it's fun to dance to, but it's not the Beatles, and it's almost four minutes long.
We started out by looking at a printout of the song lyrics, even though the lyrics are in Italian. I chose this song to discuss precisely because it's in Italian: this is a chance to think about cutting without thinking about the meaning of lyrics. For the purposes of this exercise, the voice is just another instrument.
So I asked folks to take a look at the song and - even before we listened to it - to start thinking about which parts we might consider cutting, based solely on principles like length and repetition.
f1renze and others immediately zoomed in on the fact that sections 5, 6, and 7 are repetitions of earlier sections, so perhaps we could lose one or the other of the iterations (it doesn't necessarily have to be the second one). Others noted that there's a long instrumental section between section 7 and the end, part of which could perhaps be put on the chopping block.
So how do we decide what to keep? Here we paused to listen to the song, which I am not providing in full; it's available on iTunes for $.99, and I highly recommend it, so by all means go grab it right now. I suggested that folks listen for interesting vocal effects or changes in instrumentation - any reasons to keep one section over another.
After listening, people pointed out that
Compare for yourself: section 4 vs. section 7 [right-click to save].
(You can even see these differences in the waveforms of the sounds; when the background instrumentation drops out, the drumbeats become very spiky and distinct.)
As for the concluding instrumental section, it consists of eight repetitions of an eight-beat (two or four measures, depending on how you count) pattern; cutting out half or even three-quarters of it would be easy enough.
TECHNICAL TRICKS
All this conceptual stuff is all well and good, but how about where exactly to place a cut on a waveform? I'm sure there are lots of ways of approaching this, but I can tell you what's worked for me:
So, to sum up:
I've been thinking a lot about audio editing since VividCon 2004, when several different people, talking about different vids of mine, mentioned song choice as something they thought I did well. The compliments pleased me very much, especially because song choice was in fact one of the elements I was happiest with in each of the vids in question. But they also made me chuckle a bit, because I didn't just choose those songs; I also, to some extent, built them - by which I mean that I manipulated them in various ways, from cutting whole verses to trimming phrases.
Put simply: while some songs may in fact be perfect for vidding purposes, many need a little help.
So, as I said at the beginning of the session, this was not a panel about whether to audio edit; my assumption was, and is, that audio editing is always a good thing to consider, and often a good thing to do.
NOTES ON EXAMPLES
As I said at the panel, I've drawn all examples from my own vids not because I'm the most egotistical vidder ever (...although that may well be true) but because the whole point of audio editing is that listeners shouldn't know that anything's been altered (unless, of course, they know the original song very well). So these are the audio edits I know best (and can easily export for examples). For a very recent example of extremely successful audio editing, check out sdwolfpup's "Fix You", a 2.46 vid based on a 4.55 song.
REASONS TO EDIT
- Length. If the song is over three minutes long, I'm gonna at least think about editing. I may decide not to edit after all, or the edits may not in fact bring it down to three minutes; that depends on the song and on the needs of the vid. We can all think of longer vids that work beautifully. "A Day in the Life" (by astolat and cesperanza), for example, is 4.30, and there's not a dull second in there. Those of us not vidding the Beatles, however, may need to suck it up and start snipping. Longer songs often have, let's face it, filler: repetitions of choruses, random instrumental noodling at the bridge, whatever. These elements may work for the song qua song, but a vid has to be able to do something with them. Sometimes they can be made to work; often, though, vidder and viewer alike will benefit from having them chopped out.
- Comedy. If the vid's supposed to be funny, the song needs to be even shorter. We've all seen comedy vids that were funny for the first minute and then just didn't have anywhere else to go; personally, I have never seen a comedy vid I thought was too short.
- Repetition. Sometimes the song is repeating the same lyrics, or doing the same musical thing, more times than we need.
- Lyrics. Sometimes the song is saying stuff we just can't use, period. Of course, sometimes an oblique lyric will push us to come up with a nifty metaphor or do some other clever thing, and that's cool; but I feel quite strongly that we really shouldn't be afraid to chop stuff that's going to be too much of a stretch. I've done major cuts of lyrics in two different vids (see the examples from "Window of Opportunity" and "Out Here," below), and in both cases was able to kill two birds with one stone: got rid of lyrics that I had no idea how to handle because they didn't work with the rest of the vid, and in the process reduced the song length considerably. There's no need to make ourselves jump through unnecessary hoops.
- Structure. Sometimes the song's climax is in the wrong place for the vid, or there's some other structural glitch, and we want to reshape the whole thing through editing. (Thanks to permetaform for pointing this one out during the session.)
I know several vidders brought up other excellent reasons for editing, but I didn't write them down and my memory's shot; if anyone remembers what they were, comment and I'll add them!
TOOLS FOR EDITING
Any decent video editing software (Premiere, FinalCut, etc.) will also allow for quite a lot of audio editing: snipping out chunks of the song, putting bits on different tracks and crossfading them to blend the sound if necessary. Until very recently, I did all my audio editing in Premiere - including the rather nitpicky cutting of "Thistledown Tears" (see below). As killabeez observed during the session, editing audio within the vid project has the advantage of allowing us to change our minds while vidding, to tinker with the song the same way we tinker with our clips.
There's also the option of using audio editing software; I've used the evaluation version of Goldwave, for example, and have recently started using Adobe Audition (which I love, so if you're looking for a recommendation, there you go). Several panel attendees mentioned other audio editing software they've used successfully, so ask around if you're interested. This type of software has the advantage of offering more effects (and particularly more effect presets) than video editing software. For most of us, I think any basic editor (preferably but not necessarily multitrack) will work just fine, although I do recommend making sure the software has a center channel extractor (sometimes called a "karaoke" setting) - more on this in the "Out Here" examples. Unlike Killa, I prefer to have the audio edits completed before I ever start vidding, so editing the .wav file in advance and importing that edited version into Premiere works just fine for me.
TYPES OF AUDIO EDIT
NB: These aren't technical terms at all; I'm just trying to keep terms consistent within this set of explanations.
- Fading out early. Pretty self-explanatory, and easy; works the same way a fade-to-black does for video clips, except we're adjusting volume rather than opacity. I did this with "Atropine"; the song itself fades out, so I just faded it out two minutes early.
- Trimming. Can be single chunk or multiple small snips; see examples from "Window of Opportunity" and "Thistledown Tears" below.
- Rearranging. Moving (rather than simply deleting) verses, phrases, or other chunks of the song; see examples from "Out Here" below.
MAKING SEAMLESS EDITS
Making seamless edits requires two things:
- Getting the musical pattern right. This has to do with timing-choosing the right place to splice, cutting in the right place relative to beats and actual sounds.
- Getting the sound right. This has to do with instrumentation, volume, etc.
The musical pattern part is relatively simple; we have to figure out how many measures to cut to preserve whatever musical or rhythmic line the song has established, but once we know that, it's pretty much a matter of a couple of razor cuts. (See the section on technical tricks for more about the exact how and where of the razor cuts.)
The sound part can be fairly simple or a huge pain depending on the type of music being used. Techno, for example, is extremely forgiving of audio edits; straight-ahead rock songs with a strong repetitive drumbeat and bass guitar line are generally pretty easy to edit as well. Guitar solos, however, can be a big pain in the ass, and anything with piano, strings, or noodly acoustic guitar is likely to be a bit tricky (though not impossible!). For the latter types of songs, attempts to trim and/or rearrange will probably benefit from having a multi-track editor with special effects capabilities.
TRIMMING EXAMPLE: Window of Opportunity
This song is originally 4.02; I cut it down to 3.01. That's a quarter of the song gone. Trust me when I say you don't miss it.
If you know the vid and not the original song, you know the first verse as:
I'm on vacation from the measurement of quality
backing away from sleepy hollow front lawn of tranquility
just singing in my head and wreaking havoc with authority
I've fallen out the window of opportunity
But check out the first 40 seconds of the original [right-click to save].
If you know Wonderfalls at all, you know that the lyrics of this original first verse are actually kind of terrific for Jaye, but they tell such a specific story that illustrating them with the available source was going to be challenging if not impossible. And since I knew I wanted to get the song shorter anyway, this verse got the axe straightaway.
If you count off the measures before the lyrics kick in (and pay attention to the bass line), you'll find that the song starts with a pair of four-measure patterns, which adds up to a nicely balanced eight-measure pattern. Eight measures is (as yhlee pointed out during the panel) a pretty standard musical phrase in Western music, although patterns may also come in two- or four- or sixteen-measure chunks (and, of course, in the absence of written notation, the number of measures depends on how we count, which may vary among listeners).
So one editing option would be to preserve the original eight-measure intro, and then just splice in the second verse instead of continuing with the first verse, which would sound like this:
Sample Window of Opportunity edit [right-click to save]
The problem with this strategy, to my mind, is that the two four-measure patterns of the intro sound almost exactly like each other, whereas the little four-measure musical interlude between first and second verses, though it maintains exactly the same bass and keyboard stuff, adds in drums and a bit of guitar noodling around the edges.
So, for my vid edit, I built a new intro: I used the first four measures of the original song, and then spliced in the four measures right before the second verse begins:
Actual Window of Opportunity edit [right-click to save]
Either of these edits is very simple to make: two snips, cut out the stuff in the middle, scoot the two bits up against each other, and hey presto, off and running.
TRIMMING EXAMPLE: Thistledown Tears
I didn't use this example at VividCon because we didn't have time; I'm including it here as an example of smaller-scale trimming.
"Thistledown Tears" is originally 5.13; I cut it to 3.50 (and in retrospect should probably have cut more; ah well, lessons learned). Unlike "Window," where I got rid of most of the length in a couple of large cuts, I trimmed "Thistledown" by cutting out eleven smaller chunks, some of which are only two or three measures long.
One of the things I love about "Thistledown" as a song is that it's very slightly and deliberately off-kilter in terms of its musical patterns. Where we would normally expect four- or eight-measure instrumental interludes between verses, we sometimes get five or seven measures instead. The song is also characterized by repetition of the line "night is quickly passing." Both of those elements of the song, though fun in and of themselves, are problematic in a vidding context. (Think River: "This song is problematic.") I felt that the unbalanced instrumental bits might be more distracting than interesting, and I was quite sure that, as much as I loved the repeated line and all the shades of longing and hope and weariness that Jeff brings to it, I wasn't going to be able to do something interesting with it as many times as the song included it. (Plus, he frequently eliminates it himself when he plays it live; some of the live versions end up remarkably like the edited version I made for the vid.)
Anyway... back to odd-numbered patterns and inconvenient repetitions. This snippet from the original song [right-click to save] shows both these problem elements. This snippet from the vid version shows what I did with it: cut the first iteration of "night is quickly passing," kept the first two measures of the five-measure interlude, and went right into the "hymn songs" lyric.
This bit is what I was thinking of when I said to morgandawn, during the panel, that edits don't necessarily have to maintain eight-measure patterns; in this case, we get a balanced sequence of four measures once the "extra" measure is cut out, and then I cut that pattern in half because I felt that for this particular guitar line even two measures can stand on their own as a complete musical idea: one measure of melody rising, the second falling. I did preserve the full four-measure sequence at the very beginning of the song, because it feels more complete, and I kept it again with the sequence right after the musical climax of the song, where I wanted some recovery time and also more of a bridge effect. But for interludes between verses, I decided that two measures would do - and I stuck to this throughout the song, trimming every interlude down to two measures (except at the bridge).
Because "Thistledown" is all about the acoustic guitar, the sounds are not quite as sharp and definite as they would be in more electric or electronic songs, which meant that a lot of the cuts (though not all of them) sounded better when I used tiny crossfades - and by tiny I mean "less than a second." Audition provides several different kinds of crossfade, including sinusoidal (sine wave or S-shaped) and logarithmic (a steep curve), and the logarithmic in particular can come in handy, but most of the time I find that a simple linear crossfade sounds just right. Here's a screencap of one edit and the corresponding musical snippet.
REARRANGING EXAMPLE: Out Here
It's actually rather difficult, I think, to compare versions of a song one doesn't necessarily know, especially on a single listen, so I offer instead a visual representation of the edits I made to "Out Here": a side-by-side comparison of the lyrics. Verses I cut are grayed down; the section I moved is highlighted.
Decisions about what lyrics to cut from a song will almost always be based partly on content; in this case, I cut the two verses I did largely because when I started editing I already knew that the song was going to have to reverse the meaning of the lyrics. "Out Here" is, as Peter once said, about going crazy, cutting off your hair, and moving to rural Massachusetts. For the purposes of my vid, it had to be about getting kicked out of Canada and exiled to the metaphorical wilderness of Chicago, which meant that references to actual rural locations ("got your rocks, got your rivers, got your trees") were good things to get rid of. I mean, I might have been able to make them work, but I didn't need to; the vid idea certainly didn't require it.
But decisions can't just be about content; there's also the question of structure. Each of this song's verses is in two halves that do just slightly different things musically; cut one half, and the verse sounds wrong, unbalanced. (There's also a signal in the lyrics: in the original, each of the first three verses ends with "miles away.") So in order to cut the second half of the first verse, I needed to also cut the first half of the second verse and simply smoosh the remaining halves together into a new combination. No problem.
Except I actually needed the first half of the second verse, especially the lines about "it is my privilege to lie awake at night and think of what I've lost," because if that's not Fraser in a nutshell I really don't know what is. The obvious solution was to substitute this section for the "out here it's not as bad as I tell it" section; it even worked structurally, since they're both first halves.
But when I simply spliced that section in right after the end of the guitar solo, I got something that didn't work [right-click to save]. The musical pattern's right, but the sound is all wrong: the sustained guitar note simply stops, and the sound, which has been getting fuller and louder all through the solo, drops way back.
I realized that what I needed to do was keep the first line after the guitar solo and layer it on top of the verse I was putting in its place. Except I also knew that wouldn't work, because of the lyrics. Without any adjustments, what I got was a muddy mess [right-click to save]. The guitar line does continue the way it should, but of course so does that second layer of vocals.
Clearly what I needed to do was strip out the vocals, leaving just the electric guitar and enough of the acoustic guitar to amplify the sound a bit. I poked around a bit in the tools, but couldn't find anything that was obviously what I needed; so I went to the help files, searched for "remove vocals," and discovered the Center Channel Extractor filter (sometimes called "karaoke" for obvious reasons). At the panel, absolut3destiny explained how this filter actually works; all I know is that when I applied it to my little post-guitar-solo snippet, using the male vocalist setting, I got this useful bit of sound [right-click to save].
This works. The electric guitar is coming through clearly, as are the lowest notes of the acoustic guitar; the vocals aren't entirely gone, but they're sufficiently echo-y that they aren't terribly noticeable when combined with the spliced-in verse [right-click to save]. I didn't even have to use any fades, although I would have messed around with that if the sound still seemed to need blending.
One thing I want to emphasize about this rearranging process: nothing about it was difficult. It took a while, sure, both to make the decisions about what needed doing and to figure out which of the effects at my disposal would do what I wanted; and of course there was some fiddling with settings and clip boundaries to get things to come out exactly right. But, let's face it, anybody who's not willing to take on a little bit of fiddling has probably not taken up vidding.
DISCUSSION: Tre Sorelle
In this part of the panel, I wanted to provide a hypothetical vidsong and give folks a chance to think about what they might cut if they were actually vidding it. I actually have vidded this song, and I made a particular set of choices about how to edit it, but they're not the only possible choices; there are lots of potentially right answers to the question of what to cut. There is also, however, a WRONG answer, which would be "don't cut anything". This song is techno - Italian folk-techno, granted, but still techno: it features bagpipes and violin, and it's fun to dance to, but it's not the Beatles, and it's almost four minutes long.
We started out by looking at a printout of the song lyrics, even though the lyrics are in Italian. I chose this song to discuss precisely because it's in Italian: this is a chance to think about cutting without thinking about the meaning of lyrics. For the purposes of this exercise, the voice is just another instrument.
So I asked folks to take a look at the song and - even before we listened to it - to start thinking about which parts we might consider cutting, based solely on principles like length and repetition.
f1renze and others immediately zoomed in on the fact that sections 5, 6, and 7 are repetitions of earlier sections, so perhaps we could lose one or the other of the iterations (it doesn't necessarily have to be the second one). Others noted that there's a long instrumental section between section 7 and the end, part of which could perhaps be put on the chopping block.
So how do we decide what to keep? Here we paused to listen to the song, which I am not providing in full; it's available on iTunes for $.99, and I highly recommend it, so by all means go grab it right now. I suggested that folks listen for interesting vocal effects or changes in instrumentation - any reasons to keep one section over another.
After listening, people pointed out that
- section 4 has some interesting musical differences from sections 1, 2, 3, and thus might be a good one to keep;
- section 6 repeats section 3 without significant differences;
- sections 5 and 7 both do something a little different from the corresponding sections in the first half of the song; in particular, section 7 repeats the lyrics but drops out most of the instrumentation except the drumbeat and then layers in the violin line that's filled in between verses but has not, up until this point, been combined with the vocals.
Compare for yourself: section 4 vs. section 7 [right-click to save].
(You can even see these differences in the waveforms of the sounds; when the background instrumentation drops out, the drumbeats become very spiky and distinct.)
As for the concluding instrumental section, it consists of eight repetitions of an eight-beat (two or four measures, depending on how you count) pattern; cutting out half or even three-quarters of it would be easy enough.
TECHNICAL TRICKS
All this conceptual stuff is all well and good, but how about where exactly to place a cut on a waveform? I'm sure there are lots of ways of approaching this, but I can tell you what's worked for me:
- Cut right before the peak of the waveform on the beats where you want to cut. One advantage of multitrack editors is that they make it pretty easy to see whether you've made matching cuts: pull the piece that you're cutting down into the next track and pull up the piece that's replacing it; the waveform peaks of the next few beats should line up more or less exactly (more for techno, possibly less for a song with less rigid rhythms). And, of course, use your ear and your common sense; listen to the edit, and if it sounds glitchy, fiddle with it.
- Cut on a beat, but not just any beat; make it the downbeat, and better yet make it the downbeat of the first measure of a pattern (whether that pattern is 2 or 4 or 8 measures long); this is when our ears are prepared to hear a change in the music anyway. Think back to the Window of Opportunity intro examples: if the drums kicked in after six measures (or, worse, five measures) instead of four, it would sound weird. A song determines its own balance; we need to preserve that balance as much as we can.
So, to sum up:
- It's okay to make it shorter. No, seriously, it's okay to make it shorter. [ETA: Since I gather that at least a few folks are getting bent out of shape about this opinion, possibly without reading the rest of the post, let me rephrase it a bit: Be willing to consider making it shorter.]
- Work with the song, not against it.
- Don't be afraid to cut whole verses.
- Don't be afraid to rearrange stuff.
- Playing with tech (fades, center channel extractor, etc) is fun and can be useful, but much is possible with even the most basic of tools.