SamplecrazeStretch That Note

Sampling Tools

This month’s tutorial is going to concentrate on the basic and general tools available for the sampling process and will not focus on the more detailed or esoteric tools that are adopted to further hone the sample.

So, let’s start right at the input stage of the sampler or sound card.

We have already covered the topic of attaining a clean and hot signal. Now, we need to cover the tools available to actually sample a sound, and the tools available after you have sampled a sound.

Most samplers will allow you to sample in a number of ways. But first, it is important, and sensible, to create a location for the samples. On computers, it is always good practice to create a section on your hard drive for audio. You can then create folders for your samples and have them in categories, for example, if you are sampling bass sounds, have a folder named ‘Basses’, for drums have a category named ‘Drums’ and then assign sub categories and name them relative to what you are sampling. So, for Drums, you c could have sub categories for kicks, snares and hi hats etc. This makes filing (archiving) of the samples, and even more importantly, the searching for a sample, much easier.

On hardware samplers, it is pretty much the same. You create a bank and name that and within that bank you create presets, which house the samples. On Emu samplers, the sampler creates a default preset on startup. This makes life easier. Most samplers have this facility.

Now let us look at the different ways of sampling that certain samplers provide.

I am going to concentrate on Emu Ultra samplers for this tutorial.

For the sampler to begin sampling, it needs to know a few things.

  • Source analogue or digital, 44.1 kHz or more. Pretty self explanatory as it is asking you to choose the source and the sample rate. Some samplers will have the option that will allow for digital recording as well as analogue. There are advantages to using digital recording modes but there are also disadvantages. Sample rate, however, is important. If you own a Sound Blaster card and it only operates at 48 kHz, then sampling at 44.1 kHz is not helpful at all. The other advantage of sampling at a higher rate is for precision and clarity in the representation of the sound you are sampling. Please read Part1 if any of this is confusing you. The disadvantage of higher sample rates is that they will eat up memory. In the virtual world (computer), it is now more common to sample at 24 Bits and 96 kHz (24/96) or 44.1 kHz (24/44.1). However, these parameters are dependant on the sound card you are using. If 24/96 is not supported then you cannot sample at those values.
  • Input This is for selecting mono or stereo for the sampling process.
  • Length You can predetermine the length of the sample you want to record. Maybe you only need to sample 3 seconds of a sound. Setting 3 seconds as the length automatically stops the sampler recording after 3 seconds of sampling.
  • Dither Used when recording digitally. I have explained Dither in detail in Part1, so read it again if you are still unsure of what it means.
  • Monitor Gives you the option of having it on or off. Setting it to on allows you to listen to the sound being sampled while it is being sampled.
  • Gain Here you can adjust the input gain (volume/level) of the sound (signal) being sampled. If the signal is too loud and is distorting or clipping, you can adjust the level by using this function.
  • Trigger Key This is one of the methods of sampling that I mentioned earlier. You can set the trigger key to any key on the keyboard, say C4, and when you hit C4 on your keyboard the sampler activates (gets triggered) and starts to sample.
  • Arm This puts the sampler into standby mode and when it hears a signal, it starts to sample. This is usually used in conjunction with threshold. The threshold sets the level at which you want the sampler to be triggered when in arm mode. The real advantage of this is to eliminate noise. If you set the threshold above the noise level and then play the sound, the sampler will only start to record at the threshold level setting, in this case, above the noise, as the noise is below the threshold level. The threshold/arm combination is also useful when you want to sample a sound that is above the general level of the piece of audio being sampled, an example of this would be to sample a loud snare that is above the rest of the audio piece. If you set the threshold to just below the level of the snare, the sampler ignores everything below that level and automatically records the snare.
  • Force or Manual This simply means that you press a button to start the sampler recording.

Those are the general functions available on most samplers, to do the actual sampling/recording. Now we need to look at the tools available when you press stop or complete the recording of the sound.

When you press stop, a new page appears and you are given a bunch of options. Here are the general options that are offered to you.

  • Dispose or Keep This just means you can either dump the sample, if it was no good, or keep it.
  • Place This allows you to place the sample anywhere on the keyboard you want and within this option you will have a range you can set. The range is displayed as Low and High. Let’s say I sample a C3 bass note off a synthesizer, I can then place it at C3 on my keyboard and set the low to A2 and the high to D#3. I have not placed the note and set it a range on my keyboard. This saves me loads of time and effort in having to do it later. This placing and range setting is stored in the preset, so, in effect, I am creating and building my preset as I am sampling, instead of having to sample all the notes then go back into the preset and start placing and setting ranges. Much easier. With a drum loop, you can do the same thing and by setting the range, it gives you different pitch choices of the drum loop as the sample pitches down when setting the low range value, and pitches up when setting the high range value. For single drum shots, I would place and set the ranges at the placed note. So, a kick would be placed on C1 and the low range value would be set to C1 and the same for the high range value. I now have a kick on C1.
  • Truncate Some samplers have auto-truncate and manual truncate. Truncate, also called trim or crop, is a function used to cut data before and after the sampled data. This can cut/delete space or sound before or after the sample or can be used to cut/delete any portion of the sample. Auto-truncate, simply removes everything before and after the sample.
  • NormalizeorNormalise This is a topic that has ensured some fiery debates and I doubt it will ever get resolved. Basically normalising a sample means that you raise the volume of the sample to the peak of the headroom. If, for example, you were normalising a sample to 0dB, then that means the process takes the highest peak/s in the sample data and raises them to 0dB, in this way, the loudest peak hits 0dB. This is called Peak (or absolute) normalising. By raising the highest peak you also raise the entire sample data, this has the disadvantage of raising the noise floor as well, as all data is raised till the peak hits 0dB.To normalise an audio file to ensure a certain level of perceived loudness, you need to normalise to an RMS (or relative) value of dB, rather than peak. RMS is, roughly, the average volume over a given time, rather than just the highest peak/s. It calculates the average peaks and raises those to 0dB. The disadvantage of RMS normalising is that by raising the average data peaks, you incur clipping, not always, but usually. So, in this instance, a good normalising plugin will compress or limit at the same time as normalising, so the levels do not exceed 0dB and thus, prevent clipping. I have always maintained that if you have a strong signal, with good dynamic movement, that does not clip and stays just under 0dB, then you are far better off than normalising to 0dB. Of course, there are instances where normalisation can be your friend, but in most cases, it can cause additional side effects that are not needed. These include, killing any headroom that was there, raising the noise floor so noise is also now more pronounced and evident, and to top it all off, you can get roundness in the shape of the peaks and even slight distortion or phasing. So, use this function sensibly.

Now you have your sample recorded, placed, truncated, normalised etc, you need to look at the tools available to edit and process the sample.

By selecting ‘edit sample’ you are presented with the sample and a host of tools you can use to edit and process the sample. Let us look at these briefly and then, when we come to the bigger topics, we will get a little more in depth.

  • Zoom +- This is like a magnifying tool that allows you to zoom in, or magnify, a portion of a sample.
  • Start End Size Here, you have the start and end of the sample represented in cycles and, in some samplers, in time. The size tells you the size of the sample. This might not seem important now but the size of the sample is important when working out and retuning the sample or changing the sample rate. Don’t worry, we will tackle that later.
  • Loop This is a crucial function and is the essence of what a sampler really does. The whole concept of looping is actually a simple one, whether it’s for memory saving or for creating sustained instrument sounds, the process is invaluable. What is difficult is how to find good loop points, and there are a number of reasons why this can seem complex. Firstly, unless the shape of the sample at the beginning, during and end of the loop matches up in level, shape and phase, you will have problems in finding a clean loop point. The most common enemy here is click. The best way to avoid clicks is to find what we call the ‘zero-crossing’ point. This is where the sample’s shape crosses over from the positive axis to the negative axis. At the point where the shape crosses the axis, we have a zero point. Looping at zero points eradicates the problem of clicks. But, if the shape and level don’t match up well, you will still get a click. So we are still left with a problem. What does this tell us? It tells us that the sample length being looped must be consistent, both in terms of shape and level, but also in terms of length. Too long a sample loop length and you encounter modulation. Why? Because the sample has an attack and decay. If you start your loop point too close to the attack and your end point too near the decay, you are then left with a shape that starts high and drops to a lower level, this causes the loop to modulate or wobble up and down. The opposite is also true. All sounds have a harmonic structure and if your loop length is too small then the harmonics of the sound are compromised, since you are looping a very small instance of the sample, you are, in effect, cutting the harmonics up. This will give you an unnatural loop in that it will sound very synthesized. That’s ok if you are sampling synthetic sounds but not if you are trying to loop a natural instrument sound. The final problem you are faced with is pitch. If you loop the wrong are of the sample, then it might not be in the right pitch of the original signal that was being sampled. A C3 string note will not stay at exactly C3, but move through the harmonics, so if you looped the wrong harmonic, the sample might show up as C3 +3 or worse, ie it is 3 cents off the right pitch. You need to select the most consistent part of the sample to attain the right loop points and loop length. This, unfortunately, takes practice and experience. This leads me subtly to the next function.
  • Auto Correlation Some samplers provide this function when you are looping. Basically what this function does is, after you have set your loop points, it searches for the next best loop point that it thinks will give you the best loop. Not always accurate but useful to use if you are completely off target. However, we do have another weapon at our disposal if the loop still throws up a click.
  • Crossfade Looping This technique involves fading out the end of the loop and overlapping it with a fade-in of the start of the loop, and it's a facility provided by virtually all samplers. By fading in these points, you get a smoother transition on the loop points, start and end. I only recommend using this when you have got really close to finding the right loop point and length, as it is a nice little tool and is just a polisher and not a tool to remedy bad loop points and lengths. If you had a very bad loop and it was glitching heavily, then using this tool would only make the sound unnaturally modulated, without any consistent shape. So, it’s not for error correction but for polishing off the tiny click that might be barely audible.
  • DC offset Any waveform that isn't symmetrical around the zero axis has a DC offset. DC offset is when there is too large a DC (direct current) component in the signal, sometimes visible as the signal not being visually ‘centred’ around the ‘zero level axis’. DC offsets do not affect what you actually hear, but they affect zero crossing detection and certain processing, and it is recommended that you remove them. That’s the technical, but short, definition. Basically, always remove the DC offset on a sample. This will help you find zero point crossings. This is a whole debate in itself and there are arguments raging on both sides of the fence and arguments based around the algorithms used in DC offsetting tasks. You don’t need to even think about getting involved in this debate. What you do need to know and do is to remove the DC offset on a sample and you are usually given a tool in the menu option to do this. The DC offset removal is actually called the DC filter, for those who want to know. Try experimenting, as always.

Next we will cover the methods used in sampling.