Freesound Forums

Does anyone work with loops? I just installed the app Novation Launchpad for Ipad. I have been working with loops before but then I made them in my Roland SP-404. Now I have to create my own loops in software. Audacity in my case. I’m trying launchpad as one way to make music. And it’s fun to make loops in right tempo and key. Sounds for loop making is all around you, as sounds tends to be. As usual freesound is an excellent source.
(I have three loops, I made yesterday, waiting to be accepted/moderated.)
Does anyone have any experience in loop making to share?

Well, I am not an expert loop maker, but have made a few...
Others can certainly provide additional advice, but I will make a start here.

The loops you make will be limited by two things:
1) The kind of loops you want to make (beats, bass, mellody, texture/soundscape, experimental,...)
Which, by the way also relates to what you wan to use them for.
2)What programs and gear you want to use / have access to. And also techniques to use.

I am going to assume you have some idea of what you want to do with the lopps. So really you want some ideas relating to 2), in terms of equipment and techniques.

I think the first question would be: Do you still have your SP-404? What is preventing you from making loops there and recording them as audio?

Assuming you don't have it anymore or don't want to use it. I do not know Novation Launchpad.
But I do know Audacity. I use it as a sound editor, usually not as a tool to create loops. I would usually create loops in my DAW.

I will assume Audacity is what you have or what you want to use in this case.
This is not the tool I would recommend, because the scale at the top is set to seconds and I do not think there is a way to set that to beats and set a BPM.
You can calculate beats to seconds, but maybe there are simpler ways.

Here is what I would suggest, as an easier way to create loops in Audacity:

Option 1:
Load a pre-existing loop (for example from Freesound). Preferably just a kick on the beats. Or a metronome clock would be even better.
You can adjst the tempo to what you want using the Audacity Change tempo function from the Effects menu.
Mute that track and just use the waveform as a reference to position your own samples.
Load kick, snare and other sounds and position them using the initial loop as reference for the beat.
Then select a region the same size as the initial loop you have loaded and export that to audio.

Option 2:
use a mic (cheap built in mic on your laptop will do fine) to record yourself tapping a rhythm on the desk using your fingertips. Make sure the rhythm repeats a few times.
Stop the recording. Play it back. Select the best section and play that as a loop. Edit (using cut and/or trim) to obtain a perfect loop.
Now mute it and use a similar technique as in 1, using that loop as a reference to position your samples. The difference here is that you could create the intended rhythm pattern, and all you need to do now is replace the sounds with proper ones.

These techniques work for percussion / beat loops. Creating bass or mellodic loops only from within Audacity will be considerably more difficult.

I would definetely recommend that you try using a DAW to create loops. Will be a lot easier and a million times better and more flexible.
Reaper's demo is free to download, fully functional and never expires. - You can use it at will, provided you are not making music (or loops) commercially.

Option 3:
Generate -> Click Track
(At least on my Linux box, Audacity v. 2.0.5, Click Track appears in one of the Plugins submenus of the Generate menu)

But really, loops are more naturally DAWs' business; I find LMMS a quick and easy tool to sketch out ideas, and its beat/bassline pattern editor helps with loops.

While I'm sure you are discussing musical loops, any sound can be looped, with some effort. Some things you can do in an editor can help you end up with a "better" loop, regardless of whether it is musical or not. I think most any decent editor will let you do the needed steps, but you might have to ensure that it is obeying your exact commands and not doing any "smooth transition" at the edit points, which could completely ruin your intent.

In Cool Edit Pro, I will often do this by finding general locations for where I think the sound might blend well, paying most attention to the dominant frequency of the waveform (or the most noticeable low-frequency that is modulating the amplitude) and trying to get the start and end to at least match up with respect to that aspect. I usually also try to find exact zero crossings, which can be burdensome with a stereo sound (CEP can do it automatically, but the point is that the position might shift so much that it interferes with the larger blending, so trying different locations might be necessary).

But all of that is rarely enough to get a seamless loop point, so I almost always take another step, which is to use either a section following the end or a section preceding the beginning, and cross-fade it over the other end. For sounds with complex modulations, this can create interference and cancellations, which hurts more than it helps, and in those cases it might be reasonable to use a very short segment of transition (<5ms). But otherwise I can sometimes use segments that are seconds in length, and this would be good for when the original source material was ending and there was only reverberation to follow (just use the entire tail).

So for example, if I'm using a segment after the end, then I will first use the Amplitude "Fade Out" (linear) effect on it (but I would not do this if it were merely a reverb tail, though I might want to ensure that the very end of the tail really does go to 0 amplitude), then copy it (into the cut buffer). Then at the beginning point I will "Mix Paste" (overlap) the previously-copied segment and then immediately "Undo" in order to leave the beginning piece that will be overlapped still selected in the editor. That way, I can do an Amplitude "Fade In" (linear) on that beginning piece before doing the final "Mix Paste" at exactly 100% overlap to get perfect alignment without having to count samples.

Playing through the transition in loop mode a few times, I might notice a glitch where some frequencies drop out unexpectedly due to out-of-phase cancellation at the half-way point. If so, then I know I will need to try again and to use different splice points or a shorter transition or both, so I "Undo" and try again.

Notice that in the example I just gave, I was taking the copied part from just after the end, so it is the ending splice point that is essentially repeated at the start and the end. It is that particular point that needs to be at (or very close to) a zero-crossing, because it becomes both the start and end of the resulting sound. You don't want to start a sound at a sample that is far from 0, as this creates a nasty pop or click at the onset. But the original starting point might have been a zero-crossing or not and it simply will not matter because it gets dropped to 0 via the "fade in" prior to the final overlapping paste.

This is an important point because it means you really only need a zero-crossing in the original sound at one end or the other, but not both. So whichever end you will take the cross-fade piece from will be pinned down to the zero-crossing there. To get an exact BPM, you will then need to keep the other end at the exact and precise location that allows the whole loop to have the number of samples that exactly fits the BPM you want to maintain (well, okay, you can be a few samples off, but not many!).

In any case, once done, you must have retained markers (or sample numbers) where the beginning and ending splice points were so that you can trim off all the excess. Beyond that, it is possible to include additional meta-information in the output file that you save that says that it is a loop, etc.

That's just for using a plain-old sound file as a loop. If you are setting one up for a looping synthesizer, then there are even more things to consider, as most will allow a segment at the beginning and end that are not a part of the loop itself, with the looping splice points being at offsets within the overall sound. In that case, some of the same ideas still apply to finding your looping splice points, but I really have very little experience dealing with that environment, and I don't even know if there is a standardized way to represent such things.

One caveat: If the sound to be looped has a good initial transient like a drum hit, and the final ending point is relatively quiet (very dry, little reverb), then you might get away with not even worrying about zero-crossings: just put the start and end splice points right before the initial transient of the corresponding beat, and any "glitch" from not being at a zero crossing will be so low in level compared to the initial transient that no one will ever notice it. In this respect, musical loops can often be much easier than non-musical loops.

copyc4t wrote:
Option 3:
Generate -> Click Track
(At least on my Linux box, Audacity v. 2.0.5, Click Track appears in one of the Plugins submenus of the Generate menu)

Can't believe I forgot this! Yes, this is much easier.
In the PC, "generate click track" is within the Generate menu.

@gis_sweden

How about this: a dare for everyone to create loops using Audacity?
http://www.freesound.org/forum/dare-the-community/35414/?page=1#post76235

Then everyone will be new at this because I do not think many people have used Audacity as a loop making tool before.

I still have my SP-404. You can make loops in this machine. You can export sounds as WAV. But to make smooth loops this is not the machine. And you cant resample more than 2 sounds at a time. And you cant resample a pattern you have made. You have to bounce your result to en external source and record it back into the SP. The SP is a sort of launchpad, but you cant fade in/ut sounds. There is a limitation in polyphony and processor capacity. And even the most exact cut loops will slowly drift apart. Not necessary a bad thing.
I have a Kaoss Pad 3 to. This is a nice machine when it comes to blend different sounds and create interesting effects. But you have to record sounds into the machine and record the outcome, and make a loop out of it in the computer.
That’s how I made this sound
http://www.freesound.org/people/gis_sweden/sounds/245752/

I use Audacity in windows (had Linux before but my wife didn’t like it… My kids never had a problem). And yes you have to know how long every bar is for the bpm you are working with ((60/bpm)*4) = length of a bar in seconds.
The big challenge is to make seamless loop. I tried zimbots trick a little but haven’t succeeded yet… I will practice. For a while I had a Roland W-30 sampler. Yep, then there are “even more things to consider”.

I have prepared a little track in launchpad for iPad that I will record. Then I will try to move on to more complex sounds to loop. I have some ideas… (but so little time).
Maybe the Novation Launchpad for iPad is a to “stiff” tool? I hope you understand what I mean.
Maybe one of my ideas will be in Dare 29
http://www.freesound.org/forum/dare-the-community/35414/

Thanks for your comments and ideas, freesounders!!!

(Maybe I move back to my SP to get that special SP flavor and drift apart fx…)

Earlier I used to find loops quite limited. Also it felt like plagiarising someone else's work, or cheating by using pre-made blocks to make a music piece.
I still had my mind set to the times of the Amiga sequencers, where all that you could do was take the loop and play it at different speeds.

gis_sweden wrote:
The SP is a sort of launchpad, but you cant fade in/ut sounds. There is a limitation in polyphony and processor capacity. And even the most exact cut loops will slowly drift apart. Not necessary a bad thing.

Hi Guys & Gals,

I don't make loops as such, but I do cut up and join bits of sound effects that I've recorded using
Audacity. As Zimbot said, finding zero crossing points for both channels on a stereo recording is very hit and miss, and can buggle up precise timing. My method is to find a (near) zero crossing ("C" on the keyboard in audacity). When I've selected and snipped the bit I want to loop/join I zoom right in to the first few cycles at the start, select the first half cycle of waveform and do a "Fade in" from the edit menu. I then do the same at the end of the track but with a "Fade out". You can see the start waveform change down to the zero line. Theoretically this generates a bit of distortion, bending the frequency of the first half cycle, but in practice this is so brief it is essentially inaudible. The good thing is it doesn't slow down any transient attack but it will remove any clicks at the transition points in the loop.

Another thing to check before cutting a segment for looping is to remove DC from the signal. (DC offset is something I'm noticing more these days when people send me MP4s and stuff that's been generated on phones and tablets.) A DC offset will cause Audacity to select zero crossings that aren't actually at zero but are part way up the slope of the waveform. In Audacity I remove DC by selecting the whole track and using the High-pass filter set at 1Hz. Again, on audacity's waveform display you can see the DC offset disappear.

Hope these tips are useful to somebody.

Wibby

Great tips, Wibby!

By the way, DC removal is also available as an option under Effects -> Normalize; since normalizing is an option itself, Normalize can be used just for DC removal as well.

I have tried to use ”Split Stereo Track” when working with complex waveform. Then I can treat the channels independently. When ready I select “Make Stereo Track”. Sounds like I’ve done this thousands of times…

Speaking of DC offset. Yesterday I tried to make a loop from a sound that wriggled around the 0-line and was changing in amplitude at the same time. The sound isn’t “strange” but I have not succeeded in making a loop out of it yet.

https://app.box.com/s/09hol1lenf8q5kl5wd50

I made the sound in/on Audio Sauna. I want the sound to be 4s (2 bars, 120bpm). And I want this part. I was for practice purpose, but so far I faild :-/

Oops..., no body spotted my undeliberate mistake!

In Audacity it's Z on the keyboard to find zero crossings, not C.

Copyc4t, how about that then! Yes you're right, normalize is easier to do, I never used normalize before and didn't realize it could remove DC offset without changing the gain.

Haahahaa... and I thought I was so clever working out the high-pass filter trick! I've got to go away now and think up plausible reasons why my way is much better than yours! (Well if Climate Change deniers can do it why should I concede to somebody else's superior knowledge?)

Wibby.

gis_sweden wrote:
https://app.box.com/s/09hol1lenf8q5kl5wd50

The real issue with the click is not because of the non-zero seams; you can fade in and out and still hear it, because there's a second white noise spike around 0.02, if you switch to the Spectral view in Audacity and zoom in the first half second you'll clearly see it.

If you fade-in for at least 0.05 secs you'll reduce it a lot, but it affects the overall volume too much. So you should really get rid of it at the source if you can. If you were playing a synth, raise the attack time a tiny bit and see if the problem... fades away

@gis_sweden:

how about this one? It's still better if you can correct the source though.

https://www.dropbox.com/s/1pki7v406etd8f7/basarbaudiosauna_4s-CC.wav?dl=0

@strangely_gnarled:

Your trick is clever because it's portable, you can even apply it in a live situation with a hardware filter!

Thanks copyc4t, and Harrumph.... You're even cleverer than me at thinking up my plausible self justifications!

Wibby

Yep, with your edit, the sound loops.
Hmm… I guess I was trying to change the whole envelope - afterwards. Stupid. I should go back to the source of the sound and alter the attack time.
Stubborn as I am I proceed with the same sound. Did this, and It’s a variation of zimbots trick. I copied the last second of the sound. Pasted in as new sound. Made fade in at the original sample and fade out on the 1 second part – “the new beginning”. I decreased the overall volume. Fade in/out at the ends. Adjusted the envelope. Tried “clip fix”… The sound have problems, I know… Now I will leave this sound behind me. https://app.box.com/s/3wen4scwiwe16mdww2ix

The same trick was used on this sound (maybe a dare-29 sound…)
https://app.box.com/s/b7dft8s3speb0ebgzoae

Don’t miss Dare-29! I want to see what you come up with.
http://www.freesound.org/forum/dare-the-community/35414/?page=1#post76235

I haven’t used spectral view. Scary. Sudden outbursts of energy at certain frequencies – why? – how?

gis_sweden wrote:
I haven’t used spectral view. Scary. Sudden outbursts of energy at certain frequencies – why? – how?

Low pass filter effectively extinguished the flares of energy. I actually did try low pass filter but I was to yellow to tame the scary red torch I saw. I was afraid to kill the higher frequencies…
The spikes in the ends of the sample… Strange. When you delete a part of the sample you create a red “bleeding” end of, what you call, “non-zero samples”. Also when you do some editing elsewhere , an aura of red appears…

Now this little sample has got more attention than deserved.

Now to something completely different. Or not. Some years ago I played with making tape loops from cassette tapes. That’s fun and glitchy. And you can add extra flavor with help of a magnet. Maybe it’s time to destroy a cassette again… Anybody else tried that?

gis_sweden wrote:
Also when you do some editing elsewhere , an aura of red appears…

Do you want to try something spooky? You might be tempted to process just the spike and not such a bigger part of the sound, so zoom in the spectral view and select an interval just a bit larger than the spike, then EQ or lowpass it... and you'll see what filters are made of