|
Re: Normalizing Tracks for DJ'ing
| quote: | Originally posted by daeus
I've just got hold of some CDJ's and Ive got a huge MP3 collection of 10,000+, Now I want to normalize all of these tracks because im finding the volume levels are just way out for each track to be able to actually take seriously when mixing.
Ive got hold of DBpower AMP software which was recommended by a freind, but only because he knew it had this normalization feature.
Anyway I was going to convert them all but obviously its going to take allot of space to have two copies of all the tracks ,temporarily anyway, so i was going to delete the older non converted tracks once theyve all been normalized.
Before I convert them all, any tips and DONT DO's before I take the plunge?
P.S I will be converting all to 320Kbps and I know that theyre not all 320Kbps in quality but forget that for now please as thats another headache. |
All tunes from digital record stores will peak at 0 or -0.3 for cd burning.
The tunes you find here and there on p2p are ripped of vinyl recordings and can peak at wherever. But I am not going to be an asshat and comment on how it's morally wrong to download as in most cases you can't find the tunes without paying 25-50$ at the gemm website. Let's keep the double standards for somewhere else, shall we?
Keep it real  At least I will.
If you convert a collection of mp3's that are lower than 320 to 320 you are just wasting bitrate and you are introducing the posibility of further errors in the wordlenth.
It can only sound as good as the original bitrate.
If it's 128 it's a 10 to 1 compression ratio : for every 10 it lets out 1.
If it's 192 or 256 it's a 6 to 1 compression ratio which is a little better as you actually have some distance between the lows and the highs of the tune.
If it's 320 it's a 4 to 1 compression ratio which is far better quality.
Some people claim they cant hear the difference bettween 320 and cd but really if you can't don't get into post audio you will suck.
The louder the volume the the more bits of resolution it uses.
So by compressing the fuck out of everything they can squize more sound information into little resolution. Of course the track suffocates and so does your stereo image your lower end your overtones and e.c.t. The track on mp3 will always sound squared out.
Once it's encoded you can't change that.
What I would advise you to do is simply normalize in a descent program such as sony's soundforge and save it at whatever quality it originally was.
On your custom menu select the highest encoding quality to minimize possible errors
in the wordlenth. This encoding is not the prefered method of digital record stores as it takes forever. They preffer batch encoding with a 40X speed.
If you wont use a very good program to normalize I suggest you normalize using the automatic advanced human volume sensor also known as your ears.
With one track playing in one ear you switch the cue to one track ignoring the program material coming from the speakers. Next you play a loud part of the second cd and before you even start beatmatching you match the volumes with you gains going back between the two. This is a great skill to have when playing dj vinyls in a nightclub.
This won't change the fact that some tracks will seem louder than others.
The volume will be the same but the output will be different. This is due to the fact that there is no universal standard in audio compression.
So while the volume will peak at 0 the rms value will be different.
This is how he artist and the mastering house wanted the track to sound.
That's no problem with DJ mixing.
I wouln't do any audio processing with something like ''mp3 gain''.
You will bring it down to broadcast quality if you adjust the output with like a crappy mp3 brickwall limiter.
Simply put don't fuck with the rms value of a mastered recording.
If you insist on making shit loud no matter how shitty they sound try this:
cut and paste from the recording forums:
First, let us assume you have your tracks available as one left and one right track in one or two PCM-coded files of any depth and rate.
Step one: Convert from L/R to M/S. Bring down the S content (by doing "delete" or "insert silence" or "normalize to -inf." in your editor). Convert from M/S to L/R.
Step two: Bandpass to 2k-5k. With this, you limit the energy in your music to the frequency range where the human hear is most sensitive. This not only helps you get apparent loudness, it also helps you to ruin the ears of your listeners. A lot of DJs also work along these lines!
Step Three: Normalize
Step Four: Use a brickwall limiter (actually, L2 would be ok) to limit your track. The goal here is to compensate for musicians and/or arrangers who actually built soft/loud parts and dynamics into the song. Be sure to use a short release time so you're near redline almost constantly. Thresholds of -12dB are fairly normal at this stage. Again: the trick is to set the release time really short (in the 40ms range), and be sure to turn any dithering/noiseshaping things off to not compromise the integrity of the signal in the -90dB and below range!
Step Five: Clip. Most of the stuff we have now is contained in sine waves. If you calculate the power (or energy) of your audio signal (which is similair to what the RMS value tells you), you will find that if you integrate over a sine function, you retain a 1/SQRT(2) factor. relative to optimum power efficiency. To compensate for this, you drive your signal heavily into the redline, so you get (nearly) square waves, which have the best power efficiency there is. Some "vintage" people use analogue gear for this, I'd recommend a pair of ZVEX Super-Duper-Two-In-One with both stages turned on and both gain dials (the left and right ones) cranked. Of course, use the master volume in combination with the gain of your tube DI to avoid clipping your A/D converters!
The result, and I can guarantee this, will be considerably louder than everything else you've heard before (except perhaps for some tracks by Merzbow, who realize this process as part of composing their material). Your RMS values will come close to the 0dB marker. It will sound extremely loud on almost any kind of playback equipment like MP3 players with cheap 'phones, kitchen radios, car HiFi...you name it. The only playback device it won't sound loud on would be a stand-alone subwoofer. And by the way, you can even be sure it's safe to make a record (as in vinyl) from this master!
|
