Sampling Techniques and Best Practices
I find that the most common hurdles that beginners face are that of understanding how to use their samplers, how to hook all the devices up to each other, and how to then manage the samples. The best way of tackling these sub-topics is to give you some pointers and guides, and from there, you should be able to perform the task of sampling in a coherent and ordered fashion.
Sampling is not about just recording a piece of audio, it is about organisation, management and following a protocol that ensures the best results. If these criteria are not adhered to, then you will always struggle and, more often than not, be totally disheartened by the process and results. Practice is the answer, but to be effective, one needs to follow procedure, otherwise bad habits will develop and breaking those habits becomes harder and harder with time.
Whether you are sampling in a hardware environment or software environment, the methodology is the same. You need to have a temporary location for your samples, for editing and processing, and a final destination for the samples you want to keep. For this, we have to create directories. Within those directories, we need to create sub-directories. This ensures a simple way of locating samples and makes for a neater and logical layout. So, in the case of soft sampling, ie in a computer, we need to create folders with sensible names. In the case of percussion, it makes sense to name the main folder ‘Drums’. We can then create sub-folders within the main folder and name those, for example, we could create folders with names like ‘Kicks’, ‘Snares’, ‘Hi-Hats’ and so on. We can then create another main folder and name that ‘Loops’. We can then create sub-folders and name those in accordance to BPM(Beats per minute) or genre-specific or both. An example would be ‘Hip Hop’, sub-folder ’60-85 BPM’ etc…This makes life so much easier. We can continue this method and create more folders for instrument samples or loops. You get the picture? Organisation is crucial and order is paramount. The same applies to hardware samplers. There exists, in all hardware samplers, naming and filing options. This method of archiving should be done prior to any sampling to ensure that you have a trouble-free way of following the process and retrieving the data at any time.
We now come to the path. As discussed in earlier parts of this tutorial, the signal path is the most important aspect of sampling. Keeping the signal clean and strong minimises the noise element and ensures the best dynamic range. But this is always the area that beginners struggle with. The reason for this is the lack of understanding of gain structures and the devices in the chain. Let me make that simpler to understand. Most beginners make rudimentary errors when sampling because they do not understand the nature of the sound they are sampling or the equipment being used in the signal path. The most common errors are that of recording a distorted signal, due to too high a gain, recording too low a signal, which results in adding noise when the sample is then normalised or the gain increased or encountering hum because they had to use a preamp to boost the turntable signal to be able to sample it, or when everything is absolutely right, there is still noise or hum or any artifact that cannot be traced. Of course, there are more errors than that, but these are the most basic and yet the most common, so maybe we should tackle these problems before we continue.
So, to help you understand and set up your devices a bit better the following hints and definitions will hopefully help you a tad.
1. Using a turntable
Most turntables that are stand-alone will require a preamp to boost the signal so that you can record an acceptable level. Some turntables, particularly those that are housed in hi-fi units, will have an amp built-in, but for the more pro decks, or DJ turntables, a preamp is required. The choice of a preamp is crucial. I could go into some very deep explanation about capacitance, hum, LF noise and impedance etc but that would ruin our friendship. What I will say is that the following will save you great heartache and make life a great deal easier.
Years back, the RIAA (Recording Industry Association of America) established what are known as compensation standards. The resulting RIAA preamp has been built into every hi-fi and stereo amp with phono or turntable inputs since then. In the event that you are using a turntable, connected to a mixer or stand-alone, that does not have a built-in RIAA preamp, then you would need to get one. Now, this is where the technical heads sometimes have a fiery debate. Do you apply RIAA equalisation at the preamp stage or after using software applications? Take my word for it, always apply the RIAA equalisation at the analog stage, at the preamp, and not later. This will ensure a good strong dynamic signal with ample headroom.
2. Cables
If I had a penny for every time the question of cables comes up, I would be one rich dude.
There are a few things that are crucial about cables and let us also put to bed the ridiculous analogy of ‘Expensive cables are better than cheaper cables’. This is simply not true, and if you actually took the time to make your own cables from component parts, you would realise how cheap it actually is to make your own quality cables. In fact, I will write a tutorial on this soon, along with how to build your own pop-shield. Both are crucial DIY projects that, would save you money, and are fun.
Balanced
A balanced line requires three separate conductors, two of which are signal (+ and -) and one shield/earth. You can usually determine these by looking at the connection. They will have 2 black rings and the plugs are referred to as TRS (tip, ring, and shield). Sometimes, and not always correctly, referred to as stereo jacks.
Unbalanced
An unbalanced cable runs two connectors, a hot (+) and an earth.
By the way, I am being very simplistic here as there are many variations to balanced, unbalanced, TRS, coax, etc…What is important is that if your equipment is balanced, then use balanced cables throughout the path and vice versa. The advantage of using balanced cables is one of noise reduction. Finally, if connecting balanced outputs/inputs with unbalanced cables, you can end up with signal levels that are 6dB lower than they should be. This is essentially because only half the signal is being transferred. So it always pays to match your cables.
You will find that a lot of cables are unbalanced. Guitar jack cables, speaker cables, and microphone cables being the most common.
Shielded cables can also afford better protection against RF (radio frequency) noise.
Match your cables.
Even better, switch to balanced cables, throughout the path, if possible, that way you reduce noise and cable length does not become such an issue. This has subtly led me onto the debate of length. This is, again, dependant on the type of cable and connectors. Generally, as a rule, you can use unbalanced cables with no worries at all, up to 5 metres. Balanced can go even further, 10 metres. However, these figures are not gospel.
Now we will deal with connectors. This is another area that is rife with preferences and arguments. So, I will sum up both the cable and connectors in one statement. I make my own cables but if I have to buy, then I buy Van Damme, Mogami or equivalent, and for connectors, I always use Neutrik connectors, Cannon and Switchcraft follow. My recommendation is, build your own cables. This saves money and teaches you a thing or two.
3. Ground loops, hums, power surges, and other nasty artifacts
Without going into too much detail as to what factors cause the above, I would rather propose a solution. You now have a little more insight into why certain cables can filter noise better than others, along with connectors and cable lengths and cable matching. What we now need to look at is how to prevent earth loops and surges and even hums. Most equipment needs to be earthed in some fashion and the very nature of our planet and the national grid system means we will have power surges and spikes in our mains. Add to that mains hum, or equipment hum from non-earthed equipment, and you are confronted with a multitude of problems that can all be resolved with a simple and inexpensive solution.
Nowadays, there are a number of companies that build power surge protectors in terms of mains switches, isolators for maintaining a constant predefined current, power distributors for maintaining and distributing current to a number of devices and UPS systems (uninterruptible power supply) for protection against power-downs, cuts, and outages. Simply put, you want to protect your equipment against power surges, spikes, shutdowns, etc. So, the simplest answer is to buy a power distributor that connects to all your equipment in the way of kettle plugs and sockets, a surge protector in the way of a simple mains switch breaker, found at any shop that sells plugs and the like, and that’s pretty much it.
Emo and Furman make good power distributors and protectors and they are cost-effective. Many companies make UPS systems and they can start at a very cheap bracket and go into a hefty price range, the latter being for serious users like hospitals and the like. A simple UPS system can not only protect your system against power cuts, surges, spikes but also act as a distributor for your equipment, and not break the bank either. Most commonly used when you have a computer running in your studio, and a number of other devices, that rely on a constant feed. This way, if there is a power cut in your area, the UPS will have a battery charge back-up and will continue to function, allowing you to back up your data on a computer instead of having it all wiped out by the power cut.
Personally, I have an Emo power distributor that affords me 12 kettle sockets which connect to the gear that cost me £70, and a surge protector plug set that cost me £8 from my local Maplin. If you have serious mains issues, then seek the correct help and, if possible, have an isolator specifically for your studio. If you require a UPS system, then there are a number of cheap manufacturers on the net, APC being one of the most noted. Make sure to match the power and get a True-Online type. Seek them and be happy.
Bear in mind that your turntable may cause ground hum so some type of grounding is required. With the latest Emu sound cards, notably the 1820M, there is a dedicated turntable input with a ground lug. That, to me, is one serious cost-effective way of having a sound card and a preamp with grounding, all in one unit.
4. The sound card
Probably the most confusing and wrought with obstacles is the subject of sound cards. Which one to buy, how to hook it all up, what connections, how to assign the ins and outs, analog or digital, adapt or optical, what sample rate…?
All the above can be daunting for the beginner, but it can be made easy if you understand a few very basic concepts about what the sound card is and how it functions.
As always, the goal here is to get as hot a signal as possible into the computer without noise or distortion or to compromise the headroom.
Some people like to sample digitally as opposed to analog sampling. Remember that we are in the computer’s domain here and not external hardware sampler territory. This is all about connection, so it makes sense to set your sound card’s inputs to match the incoming signal. If you are using any of the digital inputs, ADAT, SPDIF etc, then you need to select those as your inputs from the sound card’s control panel or software on the computer. If you are using the analog inputs, then you need to select these from your computer. I always recommend a hot signal at source, for example, the turntable’s preamp, after selecting the highest gain value without any distortion, you need to match the input signal by adjusting the sound card’s input gains, either from your sound card’s control panel or physically, by adjusting the trims or knobs on the sound card itself, assuming it has any. Check your meter’s in the software application that you are using to record into. Remember that in the digital domain anything above 0dB is clipping, it is not the same for the analog world, where you have some play or headroom in the signal boost. Try to keep your signal a couple of dB below 0, that way you have left enough headroom should you wish to process the sample. If you have a dead-on 0dB recording, and if you apply compression or any dynamics that boost the gain, the sample will clip. Keep it sensible.
The other area we need to touch on is the operating level.
Most pro gear operates at a nominal +4dBu and often with balanced interfaces. Most consumer or semi-pro gear uses a -10dBV operating level, and often with unbalanced interfaces. But the two levels are not interlinked or dependant. You can have +4 unbalanced or -10 balanced. These levels are measured as dBu (.775V), dBV (1V), so you can see that there is a difference in the referencing. I do not expect you to understand this as of yet, but if you want to delve into it a bit deeper, then read my Synthesis tutorials. However, you might come across certain products that are set to nominal operating levels; in this instance the gain staging is important.
5. Matching levels
It is imperative to understand how to calibrate the signal path for optimum signal to noise ratio (S/N) and to also get a true reading so that your levels show the same legending. Basically, what all this means, is that you need to be able to see the same level readouts on your hardware and software so that you are dealing with a known quantity. It is pointless if you have different gain readouts across your signal path. So, what we need to do here is to calibrate the system. In fact, it is essential to do this anyway, so that when you are mixing or producing, your levels are true. By calibrating your system and showing a true value across the path, you are then in a stronger position to be able to apply dynamics that might be dependant on numerical data as opposed to the ‘ear’ concept, that of hearing.
So, let us start at the source and finish at the destination. In this instance, the source will be the turntable, microphone or synthesizer and the destination will be the software application that you are using to sample with. For the sake of explanation, I will assume that you are using a mixer. Without a mixer, the calibration is much simpler, so I prefer to take a harder example and work off that.
The steps to follow are quite simple and make total sense.
1. Connect the source to your mixer and attain unity gain. Unity gain is a subject that is, yet again, hotly debated by tech-heads. Basically, it means to align your sound to a fader and meter readout of 0. That is very simplistic and probably means nothing to you, so I will explain in more practical terms. Let us assume that you are connecting a synthesizer to channel 1 on your mixer. You first turn the volume knob on the synthesizer to 75%, some say crank it all the way to 100%, but I prefer to leave a little room in the event that I might need to boost the signal.
Now, you set your mixer’s fader on channel 1 to 0 and the trim post or gain pots to 0. All you now need to concentrate on is the trim/gain knob. Turn this clockwise until the meter peaks at 0dB. If you do not have VU meters on your mixer, then check the LED for that channel and make sure it does not peak beyond 0dB. If you do not have an LED for individual channels, then use the master LED for the main outs, BUT make sure that every channel but channel 1 is muted. The reason for this is that ‘live’ channels will generate a certain amount of gain or noise, even if there is no signal present, and that when you sum all the channels together, then you might get a tiny amount of gain or noise at the resultant master outs. Actually, as a general rule, when you are not using a channel, mute it, this makes for a quieter mixer.
Purists will say that peaking just past 0dB is better, but that is not the case. The reason is that mixers will sum the channels to a stereo master and even if all your faders were at 0dB, the master fader could exceed the 0dB peak. For analog mixers, that is not a problem as there is ample headroom to play with. For digital mixers, that equates to clipping.
You have now achieved unity gain. Your fader is set to 0dB and your channel’s gain/trim knob controls the gain. On some mixers, you will actually see the letter U on gain/trim knobs, helping you to identify the unity location. In essence, the knob should be at U, but that is not always the case. The second method of attaining unity gain is to do the following: Mackie mixers have a U on their trim knobs, so if you set this knob to U and your fader to 0dB, then adjust the synthesizer volume till the meter peaks at 0dB, then you have attained unity gain. I have a Mackie mixer and I always end up a couple of dBs past the U setting on the trim knobs. Don’t let this worry you. What you must try to achieve is unity gain.
Ok, so we have now set unity gain for the source and the channel input on the mixer, cool. Now we need to calibrate the mixer to the sound card.
2. Now check your master outs on your mixer. I am not talking about the control room outs that are used for your monitors but the master out faders. These will be a stereo pair. A point to make here, before we carry on, is that most people will use subgroups as the outs to the sound card’s inputs. What I have done so far is to avoid the issue of subgroups or ADAT connections because I want you to understand the straight forward signal path, and that most users have a simple mixer with limited if any, subgroups.
However, treat the explanation for the master outs as if it were for the subgroup outs. At the end of the day, they are just outputs, but the beauty of subgroups is that they can be outputted to different devices and even more important, they can have different processors like gates or compressors on each subgroup, and by assigning a channel to a subgroup, you are able to have variety in your signal path. I have 8 subgroups on my mixer and I have a different compressor inserted on each one, but I have all 8 subgroups going out and into the 8 ins on my soundcard. I can then assign a number of channels to any subgroup and use any of the compressors on them, or just have 8 outs nice and clean. The other advantage of having subgroups is that you have additional EQs that you can use. Remember that the example I am giving here, of my setup, is purely for sampling purposes as I am not sampling 8 outs at the same time.
I am sampling either a mono channel or a stereo channel and the subgroups afford me further editing and processing options. For recording purposes, I would assign my subgroups differently, but we will come to that in my new tutorial about mixing and production. For now, we are only concerned with sampling.
Back on topic: Make sure your master outs are set to 0dB.
We now have unity gain from source, all the way to the destination. What you should now be getting on your meters is 0dB at channel 1 and 0dB on the master outs.
3. The sound card settings are the one area that most people have problems with. They set their sound card faders, or gain/trim knobs, at 0dB and wonder why their levels are either coming in too low or too high. If you read part 1 of this tutorial, you will understand a little more about the processes that take place within a digital domain and the A/D input stage. All you need to concern yourself with is to have unity gain right through the signal path. So, quite simply, adjust the sound card’s faders until your meters read 0dB. Open up the software application that will be doing the recording, pass a signal through the source to the destination (the application) and check the meters within the software application. There should be no, or very little, difference in the readout.
I cannot tell you how many home studios, and even pro studios, I have been to where the signal path is not calibrated and levels are all over the place. Not attaining a calibrated path results in bad mixes, confused recordings and total frustration at not being able to understand why or what is wrong with your setup.
It is also important to mention that the minute you introduce any device into this path, you will need to calibrate to compensate for the new intruder. Compressors are the real culprits here.
I will end this month’s tutorial off with a little information on the subject of noise.
Almost all devices will produce noise, all at varying degrees. Whether it is hiss, hum or just general unwanted noise, you are left with a situation whereby you want that clean signal, noise-free. The more devices you introduce into the path, the more noise is generated. Even mixers have an element of noise, generated from their circuitry. The tried and tested trick is to use noise gates or noise filters to cut out the unwanted frequencies. Some high-end mixers will have gates built into the channels for this very purpose.
You can insert a noise gate on the master outs and adjust the parameters until you eliminate the unwanted frequencies. A gate is exactly that, a gate that opens at a specified level (threshold) and shuts (release) when set to shut. You need to set the threshold to just above the noise and set the gate to stay open for infinity or a decay time that suits you. The gate will only let signals above the threshold pass through. You have parameters such as hold, release, ratio, and attack. I do not want to go into this subject in detail as I will be covering it more fully in my other tutorial, Production and Mixing. This is purely a tip to help you to maintain a clean and strong signal path.
Relevant content:
Preparing to Sample – Using hardware samplers!
Normalisation – What it is and how to use it
Topping and Tailing Ripped Beats – Truncating and Normalising
