Drum Sample Replacement with Drumagog
So let's talk about drug sample replacement. I love drum sample replacement because it allows you to kind of Ah, you can do some very, very crazy sound design when it comes to making the snare sound a certain way. And when you're dealing with real drum tracks like like the ones I have brought up here. If if I sold the snare, you're going to hear all the other drugs coming through the snare mic so that kind of that kind of limits, what I can do If I wanted to add, like, a ton of trouble to the snare, I would be increasing the trouble. The high hat as well show you right. So that kind of limits me of how far puts me in a box and gives me boundaries of how far I can push things. Eso with drum sample replacement You can just replace that that direct mike with your own sound, and it allows you to ah, do a lot more tone shaping now to do drum sample replacement. There's a couple steps to it. Um, the first thing you want to do is actually have decided which drum you're going to replace. So le...
t me show you here. I'm just gonna dragged this snare down here so we can look at it and be easier to see. So I've decided that I want to replace the snare drum. Right? And I have my edits done. Um, let me just make this one solid piece of audio, right? So the first thing you want to do is you want to detect silence to create the triggers. Explain this in a minute. So I've got I've got my original stare track and it's all edited and stuff. And now I want to I want to turn each hit into a trigger point. So you want to do that by going to detect silence? Now with the text silence, you have this control, which is a ah threshold control. So if it's all the way up, it's not going to detect any hits. But as soon as it starts to cross the snare hits, it starts to to detect hits. Now, these these different parameters you have here are going to control exactly what it what it determines as a hit. So if I if I click on small numbers here where it says minimum time open and minimum time closed, it starts to detect a lot more hits. If I have bigger numbers, it puts more time in between the hits that that it will detect. Now I've done a ton of drum replacements, so there's kind of a magic number here, and I don't even know why. But this is just the numbers that I use. So for snare, I always do. Negative. 14.7. Ah, I know it sounds strange, but that's what I've come up with is the perfect number. Pretty much works on every snare track ever. Um, and then for the minimum time, open and closed, I just make both of those 64th are 64 milliseconds. Um, it's It's very, very hard to actually hit a snare drum twice within 64 milliseconds. So the reason why you're using that number is because you don't want to. You don't want it to trigger twice on one snare hit. So when you use that number, it's kind of the golden number, and then the pre roll is just gonna be one millisecond, just so you have a little bit of audio space in front of the transient and then you don't. Your post role is gonna be milliseconds as well, so that you actually have a little bit of audio work with when you when you want to grab your edit point and I'll show you what that looks like now. So when you hit process now, every single sneer hit has been turned into a piece of audio like this. Now, the thing that happens, though, is sometimes you get like a flam hit right here. It didn't detect the second hit. So what you have to do is you have to mainly go in and cut those just so you have a, uh you want one audio event, person error hit like that and you would just literally just go through the song and mainly do that for every snare hit that it didn't detect. And this is another reason why you don't rely on computers to do stuff because they they don't have brains. Okay, so that's pretty good. Um, so let's go back to the slide. The next step is normalizing the transients. So let me show you that. Okay, so now I've got all these different snare hits, right? If I go and look at them. They're pretty much all the same velocity. But every once in a while you're going to encounter some sneer hits that, Ah, that have a lot of dynamics to them. And I'll show you Let me find a section that has a bunch of dynamics. All right, here's a good example. I'll just play this part, Okay, So some of those hair snare hits are soft, and some of them are hard. Um, so if we were to do detect silence on this part will show you what it would happen. So there's never gonna be a perfect setting where you get it, too. Pick the right snare heads like when you go and try to capture the soft hits by moving the threshold down. It captures too many hits, and if you raise it, it doesn't capture enough hits. So what you got to do is you just have to find kind of a happy balance like this, and that's kind of gonna be your hard hits. But if you're using these, these tiny little takes as trigger points. They're all going to be different volumes, and they're not gonna be consistent. So in order to make them more consistent. You use the normalized technique, which makes every single snare hit the same volume. So now if I play it, so no matter how hard he hits the drunk, it makes it the same volume. That's what the normalizing does. And that allows you to make sure that every single snare hit you captured is going to actually trigger the drum replacement, Uh, so that the sample plays properly. Okay, so you're next step after you've gotten all your snare hits cut up and you've you've normalized everything. You're gonna create velocity layers. Um, we show you how to do that. So it's gonna straight up, make four new audio tracks, and I'm gonna name that each one the velocity layer that I wanted to control. So this one's gonna be snare crack, snare, hard stare medium. They're soft, right? And we're going to assign a different pool of samples to each track. So I'm gonna go to the snare crack track, and I'm gonna load drum agog. That's what I like to use. And let's just say that we're choosing snare 11. For example, when you click on samples you're going to see all the different samples from the snare, all the different velocities. Um, since this is our crack track, we're only going to rolling, going to allow it to play crack samples. So what you do is you actually go in and click on the sample and hit solo. And by doing this every time you trigger drum agog to play a sample in that lane, it's going to play crack hit like this. So if I was to choose like a different one, like, say, the soft hit here. So even though I've got a loud sample triggering it plays back the soft one. So this is how you like manually control what each drum hit is going to what it's going to dio. Um, So if I continue doing that for the rest of the tracks, like I go into the hard one and then I go into the pool and select hard hits now I can get different snare dynamics by putting them in different lanes so I could almost, like, basically control what the drummers doing by doing by by putting it in the right lane. So that's why I like, if you were to ah, have just one type of layer. Then you would get something that sounds like this. And then if I actually use my other layer as a dynamic, that sounds a little bit more realistic. Like a drummer would actually play. Okay, so then we already went through picking the samples. Okay, so this is another major thing. You definitely want to print your replacement. I'll show you why. Let me make, like, a little part. I'm gonna change this layer. Teoh Two medium hit so we can hear the difference better. Okay, so I played it. Played it three times. I don't know if you guys could here, but there might. There was a tiny, subtle difference each time I played it. Um, turn this up a little bit. We play this back. The reason why it's different every time is because there's a pool of samples that it's selecting from. So if we watch watch this screen right here and I hit play. You're going to see these different yellow bars lighting up. So it's randomly choosing a sample from from these five that I selected its randomly choosing. Um, the reason why you want to print your sample replacement is what can happen is one of these hits could be weaker than the others. And you're going to send the song to the band and they're gonna they're gonna listen to it and be like, OK, it sounds great, but we want to change how loud the bases. So you change how loud the bases and you send it back. And now all of a sudden, one of these weaker drum hits ended up being triggered on a hit that needed to be really strong. Now they're gonna be like, Well, that hit sounds weak. So you're going to go in and try to figure out why does that part Sound week You're gonna draw Drive yourself crazy because it's gonna trigger a different sample every time. So if I like, just loop this one hit Did you hear that? That one hit was weak. So if you print your whole song, you can go through and catch all those weak hits and remove him and put in better hits, or you can just have it to where you don't select that Hit it all. You just have strong hits in your in your sample selection. So every single hit strong and then, like if you only had these two to to select from one of them would be weak. So you either want toe, limit your sample selection or print your replacement and go through and change all the ones. So we kind of did this little backwards, but I'm just gonna recap with this so looking here, we're gonna create the trigger. So you do that with detecting silence. You used the original track as a site guide, which kind of shows you what the dynamics are, and you can play off of what the drummer did in order to build your your sample replacement properly. Um, you're gonna look at what he did and try and recreate it with the different velocity layers. Uh, you with the snare drum in particular. You're gonna have ah, like four or five layers of velocity because it's a pretty dynamic instrument, and you're also gonna have four or five layers for the stair room, so I'll show you how the snare room works. So let's say you've done your whole song. You've got your whole performance done. What you would do next is you would just take all of your your snare triggers. You would select them across the entire song like this. And then you would, uh you would go control, see, to copy and then ult V to paste in place. And now that you've done that, you've got your duplicates selected on your project window. And then if you hold control and drag the top layer down toe where there's no tracks, it will create an exact copy of every hit with the exact same settings on all the tracks. And then when you go in to those tracks, you're just going to switch from it, being a direct snare two rooms there like that, and you have to redo your sample selections, Okay, And then, uh, when you playing back, you get the full snare sound. So let's take a closer look at the performance. If you look at this slide here, you can see how there's you could see those hits on the snare drum that air really loud, and you can see the really quiet hits, too, if you if you look below, you're going to see the correlation where I've lined up the hard hits with the hard snares where he was hitting this near hard, and then you can also see where I've copied his dynamics with this, the smaller heads. So it's really just a matter of matching where he hits the snare hard. You want to have a hard snare in your crack lane. And then, if he hits the snare soft, you wanna have a softer snare hit in your in your soft lane. Does that make sense? You guys have any questions about this? Have you ever seen this type of drumming placement method before? I've seen similar things, but not so in depth. The reason why I do it like this is because if you, a drama dog, has a way of taking all of these samples and sort of copying the velocity performance with these markers here like this. But what I've noticed is that a drummer is going to play certain parts differently So he might. A medium snare hit in a softer section is going to be a lot quieter than a medium snare hit in a louder section, and so that inconsistency becomes really difficult to deal with because you can't automate. I don't think you can automate thes points, so you would have section of songs where all of these things need to shift down like this because he's playing softer and you need a different Ah, you know, a different detection method and then plays louder. So now you've got to go in and raise all these up. I think they were trying to design it so that it would it would work for any kind of performance. But, um, ever since I've used drum agog, I've run into the problems when letting it decide. Like letting the computer once again letting the computer decide what to do is is always bad. So I like to take control and put the velocities where I want them to be specifically. Do you have any questions for anyone? Anyone in the chat? Oh, do you have a question? Oh, yeah, Real quick. Um, so at the end of the production and everything, generally with your work, does do the natural drum tones or anything make as much of an appearance as the triggered an engineer drum tones that you put in afterwards. Yeah, absolutely. Um, on the song that we had open earlier, the down and dirty song, I would say probably about maybe 20 to 30% of that snare sound is coming from the to snare Mike's the one that's on top in the one that's on bottom, and there's actually a little bit of ah snare mix going on there. Um, I'm mixing, Ah, a sample of room sample and then also the two mikes to get the snare sound. So, um, I would never I wouldn't say that we ever are at, like a 50 50 point. But there have been some situations like one of the Amira Records that I did the snare through. The whole record is just the rial snare. We didn't do any sample replacement because the drummer literally hit the snare so hard that it just it sounded awesome. Didn't need to be replaced. Like let's say you record five songs and you get to the sixth song and you're looking back on everything. Your recording, like All Man This Near like lost its pitch and went down, got lower because he was hitting it over and over, and it gets looser and looser so you can correct those kind of things by taking a snare hit, a snare sample that you took earlier and actually putting in drama dog you so you can actually re sample your original recording with the same drums. That's why I always tell people to take drum samples at the beginning of the session, because that's when the drums are fresh. You've got the fresh heads, they haven't been beaten to death and they're perfectly in tune. And then you go and go through and do all the songs, and you can go back and freshen up your sound by using the original samples. Awesome. Eric wants to know while watching this one question comes mine. How does it How long does it usually take for you to edit and mix a busy metal song? Like, I think the drums maybe, maybe just drums and the whole song like, How long do you spend on? Yeah, I'm sorry. Spent so some metal songs have a lot of kick drums. Um, in this song, there's a couple of sections that have a lot of double bass like this one. We played it so editing that one section would probably take a long time. Um, let me just see how long it takes me to do a couple hits because we do spend hours on this stuff, and I know that some of the editors that I work with they can usually do about three songs a day, but I don't know they might be spending that might be a 12 hour day because it might be four hours each song. So yeah, this All right, let's zoom out and see. Yeah, that was just one that's not even one bar yet. It does take a lot of time because there's thousands of drum hits and metal song, so it takes forever, Uh, and But I think every when you do that, though, it's worth it because it tighten, it tightens it up. And part of the reason why a lot of amateur productions sound bad a lot of times is timing. Yeah, I mean, quantities, quantities. ING is a huge factor in determining how polished or how slick it's going to sound. If everything was sloppy enough, time is just going to sound like a mess. Awesome. Um, Grant wants to know where you using role in triggers on each drum back in the day? How does that compare Teoh detecting silence? Yeah, that's a good question, because there's, Ah, there is a trigger that's really popular. It's a red one. It's called the D Drum Trigger. I think a lot of people start with those, and I did, too. What I found is that they break a lot. Um, and you would just have to keep buying new ones because you can't. There's no like insurance plan that you can get on so the rolling triggers airway better. They're more expensive. I think they're like between 50 to 80 bucks per trigger, but it's so much more worth it because they don't break it all. I still have the same the first that I ever bought. Still, having they still work great. It's great to have triggers because what what happens is the trigger actually receives the audio signal from the drum perfect. Like it's a 1 to 1. There's no the sound doesn't have to travel through the air to hit the mic, so it allows your edits to be extremely precise, because what you can do is you can group your triggers to your snare drunk are your snare microphones, and then when you move the trigger around, you know that you're cutting in front of the actual, the signal hitting the mic. So the signal goes from the snare drum and then to the mic. But the time that it took for that to happen is after it hit the trigger. So when you cut on the trigger, point it, it allows you to have your edits to be safe. Like I guess. Let's just wrap this up. Oh, great, let's grab us. So we've got three takeaways for the hyper manual approach that I like to take. Every single note matters and every single decision matters. This just goes back to the underlining theme by me. Having control over every single hit in which velocity layer it lands in, I can maintain the I could make sure that this song is going to sound exactly how I want it. Like one thing that you'll come across is in a chorus. If you've got a slower drumbeat, you want every single snare to be super powerful. And by letting if you let the computer control which hits its choosing, it might choose a weaker snare hit, and you want every single hit to be strong. So by doing it this this way, you can basically, uh, you convey, basically ensure that every single snare hit is exactly what you wanted to be. Train your ears and your eyes, so you need to know that. Like when you hear this kind of drum part, What would that look like? Um, I have the ability to not even look at a computer screen and listen to a drummer. Play a beat. And I know what that would automatically look like if I was looking at it on the grid. So that allows me to just kind of do things a lot faster and more efficiently, and I can get the goals that I want to achieve. Like if I'm trying to do a certain thing in the course I can. I can do it a lot easier because I know what I'm hearing. I know what that looks like, and the drum editing is very visual. Learn to listen to the drummer's. What I mean by that is you need to understand what certain fills actually our create, like how they're created. Like if a drummer hits a snare drum four times in a row, he's not gonna be able to hit the snare drum perfectly. four times, So the hardest hit is going to probably be the first hit because he's had a lot of time to swing his arm. But the next three hits he has to move his arm a lot quicker. So those those hits are gonna be quieter. And when you're when you're doing manual drummer and placement, you have the power of making every single hit as hard as you want. But if you were to do that, it wouldn't sound as good as if you were trying to make it. Match what MAWR What a human does. So when you learn to listen to drummers, you're going to learn how a drummer, actually, what he's physically capable of, and how to how to achieve that. When you're doing the manual replacement, it's important to maintain the human element of the sound. Um, and sometimes a lot of drummers air doing very strategic placement of the velocity. They might be hitting the snare drum softer first and then harder afterwards on purpose, and you want to maintain that