One of the biggest mistakes that mixing engineers make who have not dealt with a mastering room before is endeavoring to make their mixes as loud as an off-the-shelf CD. This is totally counterproductive. The final loudness of the CD is one of the issues addressed by the mastering engineer. Unfortunately many think that making the CD loud is all that mastering is. It is one of the elements of mastering, but certainly not the most important. Your source mixes should be at a relatively low level. Ideally, on a 24 bit mix the hottest peaks should hit somewhere around -6dBfs (dB full scale, meaning 6dB below zero full scale on your digital output meter). This means the average or V.U. level should be around -18 to -20dBfs.
You want to do this for two reasons: First, some of the top golden ears in our industry have determined that almost all workstations and a-to-d converters work and sound best at this 0v.u.=-18dBfs range. Whether you are mixing through an analog mixer or summing amp or in-the-box you will hear a very noticeable improvement in the clarity and space of your overall mix, especially in the vocal range. The vocals seem a lot less stressed and crowded, everything seems to breath better. Probably not coincidentally, this is the 'standard' that Sony and Phillips set for average level when they introduced the PCM digital format that we use today. The problem is it was never enforced as a true standard, so mastering and mix engineers were allowed to make CD masters as hot as they pleased, which has ruined the quality of many potentially landmark recordings since around 1995 which is when the first ridiculously loud CD's began being produced. We'll address the 'loudness wars' in a future blog entry.
The second reason you should not crowd 0dBfs on your mixes is that digital mastering processors need some headroom to work at their best. If a mix is slamming all the way to digital 0, the mastering engineer has to lower its level before he can do any digital processing, which means he's performed a dsp step before any processing has been done, and the signal quality has been slightly degraded.
The upshot of this is you should not apply any eq, compression or limiting across the stereo buss as you mix if it is for the purpose of making the mix louder.
If you mix at lower levels and without stereo buss processing as described above, the mastering engineer has his/her complete palette of tools to choose from. If the mix has not been compressed or eq'd he can choose digital, solid state analog, tube analog or any combination of processors to get the best sound possible.
Caveat: If you are using a buss compressor to get a particular compressed sound, as is often heard in heavy metal, or a rhythmic pumping you might hear in some electro-pop or dance music, and you can't get the flavor of mix you want without it, then by all means use it. But again it should be to add a particular color to the mix, not to make it loud. If you are unsure, print one mix without buss processing and another with so the mastering engineer can choose which works best for the end result.
Most problems with dynamics will usually come from drums and percussion. If you have problems with your drum tracks being too dynamic, don't compress the entire mix, assign the drums to their own subgroup and add a compressor/limiter across that group alone. This would apply to any group of instruments/vocals that you are having trouble getting to sit properly in the mix.
TWO METHODS FOR ACHIEVING -18DBFS AVERAGE LEVELS
If you are mixing through an analog console or summing amp, route a 1khz tone to the stereo output and adjust its gain so your master output V.U. meter reads 0. Then adjust the input sensitivity of the A-to-D converter that you are mixing to so its digital peak meter reads -18dBfs. Now if you mix so your output V.U. hovers around 0, you'll be hitting the digital recorder at that ideal -18dBfs average level. Some engineers prefer lower average levels of -20dBfs and even as low as -24dBfs. Experiment and listen to mixes at these different levels to learn where your A-to-D converter sounds best.
Here's a million dollar tip if you are recording and mixing in-the-box: The trick in all-digital recording and mixing is to learn to record your individual tracks at a lower level, so that during the mix you can have your channel faders close to unity gain (no boost or cut) and your master fader at unity gain, and the resultant mix be hitting that ideal 0v.u.=-18dBfs to -20dBfs range, which should put your peaks around -6dBfs. Here's the deal; the more you are boosting or cutting the level of a channel during a mix, the more dsp you are performing and the more digital artifacts you are creating. Many engineers have learned the technique of setting their monitor faders at unity gain, and then recording the individual tracks so they are getting close to an ideal mix without having to move any monitor faders up or down, and again where the average level of the monitor mix is hovering around -18dBfs. Then during the mixdown they most often have to adjust the channel faders no more than a dB or so to fine tune the mix. This takes a lot of practice to get right, as you will eventually be recording your tracks at a lower level than you are used to, but the result is a mix that has a depth and clarity to it that you've probably not achieved before.
MAJOR CAVEAT: The above only applies if you are tracking at 24 bit resolution. By employing this technique some of your tracks may be recorded at a very low level. The digital noise floor of a 24 bit recording is so far down that this isn't an issue, but a 16 bit recording is a different matter, and if you record low level tracks at 16 bit you will probably have a hashy mess when you mix. If you are tracking at 16 bits you should endeavor to record each track within 2 or 3dB of 0 full scale.
You should not worry about the perceived volumes of your songs being different. Again one of the jobs of the mastering engineer is to match the perceived levels as well as eq so there is a consistency to the tracks in both volume and tonal balance.
Many engineers have already learned that mixing at the proper levels for mastering can potentially cause a problem: If they give their clients a reference CD of the mixes that is half the volume of an off-the-shelf CD or lower, then the client will likely freak out. So now it is common practice for many engineers to do proper 24 bit high resolution, low level mixes to be delivered for mastering, and then add a limiter across the stereo buss, crank up the level and run the mixes again for making the client's reference disc.
No comments:
Post a Comment