This article was originally writting in December 2000 and deals with the problem of playing live. It argues that using MP3's ain't that bad (at that time this was still a big 'no-no' for part organizers) and allthough the article starts to feel old, we leave it here as a 'blast from the past'. Enjoy it.
This page covers some comments and opinions on the basic skills and attitude of a DJ. How does a typical audience behave ? What they expect and how they react. Afterward we give a set of recurring, but wrong, ideas about 'DJ-ing'.
The audience is the most important of all. They are the final judge of what you're doing. A party without people is no party. It is as simple as that. People can be at a party for a number of reasons
So, a party is not that difficult at all. Nevertheless some DJ's have quite a strange picture of what's going on at parties.
Dirty Trick: If you throw salt over the floor before people come in then the audience will drink more later on the evening.
A number of DJ's have the tendency to act as if they are god. This is wrong, they are not. A number of these wrong attitudes are stated below.
And the worst kind of god-attitude that is embodied within current day DJ's and Party-organizers is that:
It is just required to point out that more and more DJ's nowadays use CD's, which weren't cool enough a while ago, then we can safely say that this kind of 'oops, I don't like it when the world changes'-attitude won't bring them any further.
On top of these strange attitudes, we have DJ's who wants to use certain drugs to play. They probably think it is cool to be high (or down, or slow or fast). I wouldn't advice to behave like this: stay sober. The essence of whatever you are doing is to feel good. If you don't you will not have enough focus to do what you are payed for. If you want to gain some confidence, be fit! Do exercises every day and eat well (healthy food). It really helps to do some exercises before I go to a club (even if it's only DJ's meeting not a party). But, let's skip this 'how-fucked-up-is-the-DJ-exactly' crap. Let's continue with the real work:
Whether you have CD's, vinyl or MP3's, have an index at hand, sorted by style, annotate with the BPM and marked with the 'sound-color'. This list should contain cross references between styles: 'switch to this style using this song'. On top of this style list, also have a full index by name available every time you play.
Creating such a list takes a lot of time. You can easily spend months to create it, but when you have such a list it is your treasure. This will be half the money you make with DJ'ing. So never give this away.
Warning: A mistake often made is only to exploit the index and not to explore anymore. This is wrong because you might encounter better mixes. It's also wrong because you're definitely not looking at the audience, and above all it's wrong because DJ'ing isn't fun anymore this way.
It is also a very good idea to accurately measure the tempo of all your songs (that is up to 1/100 of a BPM). Programs such as BpmDj, BpmCount or BpmLive can help you with this. The tempo in general is necessary to a) match the tempo of the new song to the old song and b) set the tempo of various effect boxes exactly to the current playing tempo.
Tip #1: Comb the aisles of your local record store, hit all the online music outlets and follow the dots from song to song and artist to artist (this is the fun part of DJ'ing - research).
Tip #2: ask producers for previews of music and songs that will come out.
The sound quality, timbre, color is created by the original artist. A DJ cannot modify this much. Your set on the other hand belongs to you and the artist has little control over it. What audience you will see depends on the party organizers.
This means that if the songs is not good or boring that it is not your responsibility to fix it. You should select songs that are already good in the first place. How you weave them together in your set is on the other hand your task. Even 5 minutes of crappy songs can ruin your set, so be sure to use the best music you find.
Often if you buy a new CD you want to know what is usable on that CD. In general if I get a new CD I play it from front to back. If I survive that first play then it is a good record. If not then hopefully there were some excellent and remarkable good songs in it. Once this is done, I'm interested in finding mixes that fits to the good songs. This requires some fiddling. Which song can be linked to this cool song. Basic trial and error. I do however only work with the songs that are worth it. I don't spend much time on songs that 'might be good if handled like this or this'. Bottom-line is that these were not sufficiently good in the first place.
A good strategy to play music for a specific audience is to rely on a number of prototype people you know that like the music that is typically played at a certain kind of party. Think: 'would this person like this music ?' This works quite well !
Now (halfway the guide), let's start with the basics. The mixing table. A normal mixing table has a number of mono and stereo channels. We are only interested in stereo channels. Every channel has
Aside from all these things for every channel we also tend to find the following
So, when we want to throw in a new song (song B),
Technically this is not difficult at all. However, this scheme should be remembered very accurately and practice is necessary. Otherwise, one of the steps is easily looked over. Especially checking the correct gain is important because it avoids 'not-loud-enough, so let turn up the volume a bit more, oops can't go further...'-problems.
If you want to play MP3's, be sure to have spare parts (an extra computer for instance) at hand. One PC with two sound cards is possible but you still have your single point of failure. Use a good sound card, with a good sound quality.
Have all music on both PC's (if one fails you can continue with the second PC), connect them with a small hub if you want. Also, have CD's at hand to continue when both PC's fail.
When playing MP3's, using Winamp
When playing MP3's, using Linux (which would be the more sensible choice given its stability)
Some of the current day digital sound cards (like the Ensoniq chips which can be found in all new sound blaster cards), have the tendency to increase the level of your signal higher than possible. This result in digital distortion which is at least ugly. To avoid this don't set your PCM-volume to 100% but place it at 80% (or even less).
If you want to do beat mixing you need a way to change the tempo of your music. Since this is done by stretching or shrinking the time the music plays, the music will also change in pitch. If we stretch the music the pitch will go lower, if we shrink the music the pitch will rise. The question now is whether the standard pitch shifting software is good enough ? We need BPM's with an accuracy of 0.05 BPM. At 160BPM this requires us the ability to shift the pitch with 0.0003125 %. At 120BPM this requires that we have the ability to shift the pitch with 0.00041666... % The best pitch shifter I encountered till now is AlsaPlayer. This program can shift the pitch of music with 0.01% with is definitely not good enough !
Now, even if they are not accurately enough, it is possible you want to use things like AlsaPlayer (it is good software after all). If that is the case, I want to warn for one thing. Suppose we have song B at tempo 140 and song A at tempo 135. We want to match those two songs. Therefore we calculate the pitch-shift, but instead of dividing 135 by 140, we divide 140 by 135. We see a number (1.03703...) larger than one, while we expected a number lesser than one. As such we take the difference with one (0.03703...) and subtract it from 1, which gives 0.962962... (Not so) Strangely, this is another result than 135 divided by 140, which is 0.964285.
When playing different kinds of popular music, the most important is to know what is popular with the audience. On top of this there are a number of rules.
Switching from song A to B
At every moment have a list of the three/four/five next songs you will play, this should ensure continuity. If people ask something, don't switch immediately, put them at the end of your list, and eventually adapt your list. Trusts people's opinion only when they are happy. Otherwise neglect them.
Don't play killer music. Killer music is music where you loose a lot of people. For example. If you have 32 diagonal spread, seriously drunk people with 16 man and 16 woman, at 6:00 'o clock in the morning do not play a slow. They will go home afterward. (OK maybe that was the intention :). Another example is a blues-party (Hmm... no point in partying if your girlfriend has left, you are broke, your car has been stolen and you are constipated... Still wondering how fucked up people are when they go to blues-'parties'.); if you play at 3 o'clock a Techno-song (even if it is a good one), they will throw beer at you and drag you away. Knowing what an audience likes is as important as knowing what the audience absolutely dislikes. You don't want to play a killer song, not even by accident.
Essentially, it is much easier to play the same style (Techno, House, Acid, and so on) whole night than playing different styles. Of course, you have to know the style before you even think of playing. E.g: don't play Salsa if you don't know shit about it.
Now, something more difficult: Beat Mixing. Beat mixing is mixing two beats exactly over each other during a certain period. The difficulty with this is that different songs have different tempos. In the upcoming discussion we refer to song B as the one which will be mixed over song A. Synchronizing B with A is the first problem, keeping them synchronized is the second. This discussion is aimed at MP3-players.
In general beat-mixing is only possible when the two songs are
playing at the same speed. Therefore, one needs to bring the tempo
of one of both songs to the tempo of the other song. However: knowing the
tempo of a song up to an accuracy of 1BPM is not even enough to keep
two songs synchronized over 1 measure. How accurately the tempo of both
songs needs to be known is discussed
here. Generally, you will need to stay in touch with both songs while they are playing.
This however forms a problem because
Therefore , during playing one needs the ability to shift a playing song a bit forward or a bit backward, such that they stay synchronized. This is called nudging. A nudge typically consists of shifting the song 5 to 10 ms. This is around 1/64 note.
When a suitable song has been selected and it is playing at the correct tempo one needs to start the song at the correct moment. Typically this moment is at the beginning of a phrase (that is the beginning of 8 measures). Normally, when the song is started it won't start exactly at the moment you intended it. Therefore, you will need to nudge a little bit. This however is not easy because it is difficult to decide whether the song you threw in started too late or too early. For instance, in the figure below, the white line is the time-line of the main song. The red line is the monitor song which has been started too late. The blue line is the same monitor song but started too early. As can be seen, if we only listen to the beats, it is impossible to distinguish whether the song is too late or too early.
Nevertheless, we do not necessarily need to listen only to the bass-drums, we can also listen to the entire song. This however is also a problem because it becomes simply a chaotic piece of audio which is very difficult to interpret consciously. However, unconsciously it is possible to hear the difference. Therefore, one only needs to try to follow the music and focus specially on one of the both songs. The song for which it is easy to differentiate it from the rest and keep on focusing on it is the first song. For instance, in the red case, the song which can relatively easily be listened to is the white one, our main song. Hence, the monitor song comes too late. In the blue case, we will easily focus on the blue song, the monitor song, hence the monitor songs comes too early.
Another pragmatic way to solve this problem is to nudge forward, if the problem becomes worse, nudge two times backward.
During the time the two songs overlap the tempo difference between the two songs (even if it is a very small tempo difference) will result in a slight synchronization drift. This is pictured in the figure below
To solve this one needs to know beforehand which song is the slowest one of both. before a mix is done. Solve this problem is easy. Make sure both song are synchronized, now wait until the two beats sound double. Nudge forward. If it becomes better, you should keep on nudging forward since the second song is going a bit too slow. If it becomes worse you should nudge 2 times backward and conclude that the second song is going a bit too fast. To be workable a DJ should maximally nudge every 4 beats, otherwise he has simply a wrong tempo and should change the tempo of one of both. The direction determined by this technique is the direction you need to use to keep them synchronized once they have been synchronized.
When you finally have the two beats exactly over each other in your headphones you want to switch slowly to song B. Before you do this be sure to cut off the bass drum with the equalizer. Otherwise you get a very ugly flanging effect on the bass drums. If the volume is good, switch off song A's bass drum while you turn on songs B's bass drum. This way it will go unnoticed. If you need to nudge to keep the tempo up during fading, watch the hi hats, not the bass drum, you won't hear it.
Most songs are in a 4/4 rhythm and it is in general a good idea to respect this pattern: multiples of 4. 4 beats in one measure & 4 (or 8) measures in a sequence. If you respect this you will find that you get easily into the flow of mixing. Of course, this requires some practice, but after a while you will actually start using this scheme. Each 4 measures you can change something like cutting away the bass-drum of one song in favor of the other or using the 4 beats/4 measures knowledge to add breaks and gaps in the music at appropriate places. Such breaks will also ensure that the audience does not loose track of the underlying synchronization.
Once you have learned how to cross-fade two songs, you might want to experiment with sudden breaks and gaps in the music. This will give the music more punch and keep people dancing.
Sound effects are nice things since they make the life of the DJ entertaining. There are a plethora of sound effects out there and probably even more effect boxes. In most cases they can be divided in broad categories that reflect the underlying signal processing algorithms. E.g, LFO (Low frequency oscillators), flangers, phasers, loops, reverbs etc. When DJ'ing techno and electronic music such effects can help to spice up your set. The key to using effects is first to know your hardware and setup (the wiring to/from the mixing desk/effect box(es)). Since there are so many possible routing setups it is probably useful to stick to a common routing which is the sound signal that goes to the effect box, which in turn will alter the sound and return it to the mixing desk. There the 'wet' (affected/effected signal) is mixed again with the dry (unaffected) signal. When the effect is over it is often useful to fade out the effect instead of throwing out instantaneously. For hard effects this is necessary to avoid too sudden changes.
A usable strategy to learn to use effect boxes is a) practice, b) record, c) keep a chart handy of what works and when it works. You'll find that you might tend to favor specific effects when you are playing but when you re-listen you'll notice that you might have overdone it, or placed the effect at the wrong place. Below I list my experience with effects
Are an excellent too to add ambiance to certain passages. It can however become somewhat too dense in tracks that require punch.
This is a kind of on the fly within measure resampling of a sampled loop. This often leads to 'matrix' kind of effects.
Is an effect where the current playing song is sampled and immediately replaced by a looping version of the short sampled fragment. The result can be useful to build up bassdrum rolls or snare rolls even if that was not present in the original song. The effect is often very invasive, and can be combined with digital filters or flangers. In all cases, the use of this effect should be minimized to once every hour or so, otherwise your set might become transparent
Is an effect that will move the music from left to right in an automatic fashion. Depending on the number of autopanners and whether they are tempo sensitive or not, a number of interesting things can be done
Having seen the Goa scene adopt/embrace ridiculously stupid samples I can only comment that samples when used extremely spare (1 voice sample on an entire CD), will make that people remember that song, which is exactly the reason why producers insist on some shitty voice sample. Let me repeat that:
This of course leads producers to suggest that each song should contain some voice sample. However, in the end this works against the song since movie samples and voice samples without sensible meaning are plain stupid.
Delays are different from reverbs. While reverbs tend to be impulse expanders, delays will copy the signal to itself. From a technical point of view the implementations are completely different. To understand this, try to create a reverb with a delay effect by shortening the delay time. You'll notice that you are actually creating a digital bandpass filter the shorter your delay time becomes. Reverbs don't have this problem. However, it is somewhat more difficult to create reverbs that go on for long times. In any case.
... add space to stuff. Euhm. Well, yes they add space in general.
Are multirate filter banks that will split the signal in a large number of subsignals, that if combined will produce the original sound again. Of course, it is possible to choose which part of the signal should be reused in the end. These tools are used to have the effect of a 'talking synth', which is a good effect for certain songs and to convey a kind of robot life. Producers nowadays also use this effect to fix the problem of singers that cannot stick to their tone. This gave the very unnatural singing effect and to be honest -> I hate it. If there is a dance song with a vocoder in it I turn it off, mainly because I know that the effect is used for people that cannot sing.
A good tip to get in touch with producers and managers that can get your career started is to try to find producers for new hot tracks. If you can connect to them you might get previews of new interesting tracks, which can help your own career forward as well.
A consideration when multiple DJ's are playing is to know your position and in general the main aim should be to preserve the flow of the previous DJ and not make the next DJ his life overly complicated. In a sense, the previous DJ decides what your first half hour will be since you will need to shift from their style to yours.
What you play also depends on the time of night. If you're the star playing at 01:00 (or midnight), you'll probably play some killer tunes to shake that bodies, while if you play at 10:00 you might want to play anything you want :-)
To achieve a nice take over it is also valuable to chat with the previous DJ to check out what style he/she intends to play. obviously that can change, but having some idea can both make you more prepared to jump in, and also be nice to the next DJ but not making your final song something that completely clashes with their style.
In general, one should not be too concerned with the overall layout of the night because this is the responsibility of the party organizers, who should know what you play and planned the night accordingly.
From the mailing list: As for how I play the night, I have a couple of discs of songs that I know will get people on the floor. I plan the night to flow like a roller-coaster - people on the floor all night looks good, but if you're playing a club you want them to be going to the bar and buying drinks too. I've had a few times when I've played something I thought would keep people on the floor, only to lose them and needed to bring them back with a crowd-pleaser.
The length of your set determines what to can play. If you play for an hour you might want to stick to a limited number of styles. If you play the whole night you might want to alternate styles once in a while and if you play an hour or two hours you can consider your set to be a form of showcase of your capabilities
A thing I noticed with many DJ's is that they often mix too fast and somewhat chaotic during the first 20 minutes of their set. As DJ you will be nervous the first 20 minutes and there is little you can do about it, except to know this and accept it as a fact of life and get over it. Such stage fear is normal and the best thing to do is to breath in and focus on the job at hand and force time to be steady. That is: measure how long you play each song and make sure it is not 40 seconds or so :-) Of course it is pretty normal to make this mistake because you might ask yourself every second: 'how is it going, should I change tracks, style ?'. By the end of the minute you will certainly have decided to change both track and style. Clearly a bad plan. Just relax. And by consciously relaxing you might find that it becomes easier to relax. Well... This starts to sound like some hypnotic regression analysis. In any case, observe the stress, accept it and let it go. That is the trick.
Then of course there is the opposite problem that some DJ's exhibit. After an hour or so they left the stress completely behind and are at the same stress level as the audience. This is also a bad plan since the audience is there to relax while you are there to work. So there should be a certain stress difference between you and the audience. What I noticed is that DJ's that have fun and laugh and tell jokes often don't mix that well, while those that focus and concentrate on their set do a much better job, although they might not show the smile everybody is waiting for :-)
Of course another source of this problem might be that these DJ's practice their mixing skills on sets of only half an hour or so. In that case they run out of juice after an hour and certainly after 1.5 hour. If you practice: make sure your sets are sufficiently long.
Listen to your own mixes after you made them and see what can be done better. On a side note here: something I picked up from educational psychology is to always stop practicing after you performed something good. In this case, after a good beatmix, stop your session. If you do this after a bad mix because you are frustrated then you will not learn new skills.
Two other problems taken from the mailing list: