DI Critical Mass
Join Date: Nov 2005
Location: Dancing with wolves
There's not a huge amount more to go into that isn't either software/hardware specific or down to the kind of personal preferances developed through learning and experimenting by yourself. mostly tips and tricks. there are a couple of big areas that can have a little light shed upon 'em though - one of 'em is how remixes and mashups are made.
the fundamental, magic secret to remixing a tune or making a mashup of two different tunes is getting hold of the separate vocal (known as acapella), instrumental and even individual unmixed tracks. although you can work with samples taken from the finished commercially available mix of a tune that is basically nothing more than using some clever sampling - plenty of practise will soon allow you to work out what does or doesn't work there.
The remixes you hear released as single B-sides and chart dance versions of commercial songs tend to be made at the invite of the producer to bump up sales by catching the attention of a market who would otherwise be uninterested - although sometimes a bedroom producer cn sample a song and get sample clearance from the original artist resulting in a breakthrough hit, usually it is producers with established industry contacts who get sent the unmixed original audio tracks and are asked to produce a remix.
Many of the mashups made famous by the likes of 2 many DJs or Coldcut involve taking the acapella/beats/instrumental/bassline etc. from one song and placing it over parts of another song, preferably a comically unlikely choice - the skill reqiuired to do this well is often underestimated, there's a knack for hearing what works.
Unless you happen to be a very big name who gets invites from record labels to work with their tunes or you have real good contacts within the industry, you will need another source of unmixed material to work with. good sources are:
- B-sides of old singles, and more modern hiphop, RnB and Soul stuff - especially on vinyls. acapellas and instrumental tracks are often found here released specifically for DJs to work with.
- Internet search for Royalty Free Acapellas and Samples - even a quick google will do wonders.
- Remix competition files: probably the only time you'll get a chance to play with the kind of unmixed broken down tracks that industry producers are invited to work with.
- I've even found stuff in really random places like recordings from SNL appearances available on BitTorrent sites. it's like looking for good samples or VJ material - you never ever stop, 24 hours a day you're thinking "that would work well with X..."
Once you have your source material, you need to use your audio editing skills to manipulate it into the wonderful idea floating round your head. much of the basic preparation can be done using techniques described above under "Audio Editing". you will need to select loops and sections of audio to use, trim them to length and make sure they loop right. then comes the part that requires some skills.
to be honest, first i'd put the different bits in a sequencer with a metronome or constant 4/4 snare beat and just listen to see if the tempo drifts. if not, all you have to do is calculate the tempo and key of the loops and sections.
you can use some software that automatically detects tempo (good sequencers like cubase will do this for each audio sample imported) or whack it into mixmeister (or you could use that software some dude keeps advertising round here - "mixed in key") to see what the tempo and key are.
if you don't have auto dectecting software, key is easy to find - just play a constant synth not along with the sample and find the note that is in tune with it - it's simply a matter of trusting your ears (this way is often better because you can find non-standard unings to match bits of audio that some programs wouldn't give you).
likewise, tempo is easy to calculate yourself - just loop the sample in a sequencer and adjust the tempo till the loop plays back perfectly - that's the tempo of your sample.
if the tempo varies 'cos it's not an electronically secquenced modern rigid tune, then you've got a variety of ways to work around it.
the most common, and you'll notice the one that tends to be used by most pro remixers and producers is to simply take the few bars containing the hook and use that as a sample which you loop as necessary - the rest you recreate your self with as similar or as different instruments as you choose. consider madonna's recent tune where the producer (that bloke from Les Rhythmes Digitales i think) took a very small snippet of the musical hook from Abba's tune and looped it, but used his own disco and electronic bits to make the rest of the tune.
you can as an alternative settle on a base tempo to work from (this is where the use of programs like mixmeister comes in handy) and split the original tune into sections, short enough that the tempo variation is either not noticable or that break on sudden changes. then timestretch them (i'll discuss this more below) to all match your base tempo. this is slightly time consuming and only really works if there's changes from verse to chorus and back, rather than a drift across the course of the piece.
as a more advanced solution, sequencers like cubase and the more recent versions of audio editors like soundforge allow quantize functions to be applied to an audio file. this requires some serious knowledge and tinkering though 'cos the automatic quantize point etection or equivalents never seem to do as good a job as you would like for something with as rigid a tempo as most 4/4 EDM.
i like to settle on a combination of these methods, selecting relatively short blocks of bars - maybe a chorus or musical hook, basically to be treated as samples. i then calcualte an accurate tempo to the nearest decimal place for each one and timestretch them all to match each other, or more often to a nice 90 - 180bpm depending on the end genre i intend to turn the tune into.
I would actually avoid performing any of this tempo manipulation until you have calculated the tempo of all the different audio parts you want to use and settled on the tempo for the final piece - the less you fool with the samples the better as you will lose data and hence quality of sound.
Once you have settled on a tempo to work with (usually either the tempo of the vocal - this will sound most odd when timestretched too much, or the tempo defined by the genre of the piece e.g. say 128 - 140 bpm for trance maybe, or 165 - 190 bpm for DnB remembering that a sample at 90 bpm needs little editing to fit DnB - the beat at double the tempo will just sit on top!) you will then need to manipulate each sample to match that tempo. this is done either by a straight speed change (uncommon since this affects pitch, resulting in chipmunk vocals or wooly basslines, but it can work well - DnB relies on the pitchshift of the sped up drums for it's distinctive sound) or timestretching. Although it is possible to timestretch without a piece of software that has that function, i'm not even going into that here - it's time consuming and generally has poor results, i reserve that for special occasions when i want a very particular effect.
Timestretch is a function that is present in most good sequencers such as cubase (i can't say for certain which do and don't - i'm not familiar with 'em all) and good audio editors such as soundforge. It allows you to alter the tempo/length of a piece of audio without changing the pitch. usually it has a few basic features such as the option to input the original tempo of the sample, the desired output tempo and the type of audio e.g. drum, vocal etc. there will often be many more parameters you can control, but these are for more advanced discussions.
For now i would settle for just ensuring that you select the best algorhythm for the program to use - i.e. if you're using a drum loop then tell it drums or select timing accuracy over sound. if you're using a vocal loop then tell it vocal or select quality rather than accurate timing. there's a bit of a skill to getting the best results - options like minimum flanging or echo, or high quality all have their advantages but aren't necessarily the best option.
it is definitely true that some timestretch programs and functions are better than others even when used well - soundforge can produce much better qulaity results than cubase and manage a greater ratio - i.e. a bigger increase or decrease in tempo/length. on the other hand, if you want that weird stretched out metallic groaning noise that is sometimes very atmospheric, cubase has an algorhythm that does it great!
to continue giving instructions in the wrong order, it has just struck me that it is generally better to deal with pitch/key of a sample before tempo/length. most pitchshift algorhythms that leave tempo the same have poor results quality wise compared to those that result in a new tempo. i often change pitch first then calculate the tempo and timestretch the sample to the desired final tempo.
Although programs like mixmeister will tell you the key of a song, loop or sample, and hence give you an idea of how much pitchshift is required to make the whole lot sound good together, even if you choose not to make 'em the same key, you don't have to follow the rules of "camelot notation" (search forums for harmonic mixing/camelot notation). whatever sounds good to you is fine - it's your music. but generally, you're gonna want the vocals/instrumentals in complementary keys if not the same one.
once you've got all the different sections, loops and samples in the right key and tempo, you've just got to sequence them - if you're making a mashup then it's justa case of throwing together whaterver bits sound good.
if you're doing a reimx, then you're gonna be using as much of your own material as you choose to add/alter the sound to produce a new song in a different genre/length/mood or whatever.
all of this is a hell of a lot simpler if you're using a nice recent sequencer such as Cubase SX3, cos you can calculate tempo, perform near lossless constant pitch time stretching and do all the other fooling around with your audio files in one quick process.
for those who have never used such high end sequencers, they may seem bewilderingly complex but to be able to simply select an audio track, import a sample, automatically create a new version to work with, adjust tempo and pitch, then quickly move it and copy it as necessary, trimming start and end points where needed all in a few clicks is very convenient and allows you to concentrate on getting the sounds in your head "down on to paper" rather than battling with the mechanics of audio work.
all i do outside of cubase is use soundforge to sample, find the zero-crossing point and make ther perfect loops for the original audiofiles!
"It sounds like God treading on a lego brick" - Jeremy Clarkson
New Psy-Dub remix!