Tex - - That is exactly what I said earlier, and why I have been reluctant to put some real "meat" on the problem. Whatever I would say - some people would disagree - and some (most) wouldn't understand - and by saying I'm not giving details some would accuse me of being arrogant.
The fact is, this is a developing technology. Things are being discovered every day. The reason sound is so difficult is that sound itself is not technology, its art, its a feel. Therefore the computing process involved is more complex, the data processing more complex.
Its like converters - everyone talks about them all the time, everyone has got an opinion on them, loads of people can tell you which chip is in which converter, the dB headroom etc. from 'homies' to seasoned pros........... ok cool, and what about the analogue modeling which can make 2 converters with the same tech spec sound entirely different - indeed can make one sound like shit and the other good? Never discussed, because it’s not known on the 'outside', but it’s perhaps the most important aspect of a converter.
There are many such elements in digital audio. And some of them we find very difficult to understand. We have a "thing" going on with a SRC192 sample rate converter at the moment. Its clocked independently at high precision, it does its job extremely well, but as it was in the chain already I left it on as the last link when I had to burn a CD from a file already at 16/44.1 - result? Surprising improvement in sound quality on the CD. Why? God knows, well I hope he does, because I don't and a whole team of engineers who designed the SRC don't know either, the only thing we know is that somehow the SRC 're-aligns clock and audio prior to leaving the unit', which makes it sound good.
So we are back to what I stated many times before and got slated for many times, which are all the usual phrases people hate to hear:
- You get what you pay for
- Do as good as you can with the gear you have got
- Spend your money wisely to improve your SYSTEM first, before you spend money on fancy preamps, effect processors, microphones, otherwise ... if it goes to disk in bad quality ........
- We can now get the same results at home as we can in a pro studio --- yeah right.
The simple fact is that summing is a problem mainly caused by fixed point processing.
Would you like to eliminate fixed point processing and change to floating point processing? Yeah, no doubt. Go and buy one of the mainframe consoles out there - you can get some good ones for a couple of hundreds of thousands (prices dropping monthly!)
(note: there are also some pretty bad design cheaper ones on the market)
In the meantime, you work on a DAW - you are stuck with fixed point processing. Some are good for what they are (Pro Tools, SAW), some are not so good, and most are downright lousy.
One thing for sure, they keep improving.
Like Tex suggested, in the meantime lets stick to reality and work with what we've got. So........ here are some "real world" do's and don't, what issis and what isntissis
- If you record through a DAW with fixed point processing the optimum level is unity, 0dB. That is where you get the maximum bit rate your system has been set-up for (16 or 24 bit). So, if you record at 16 bit, unity is 16 bit, at 24 bit, unity is 24 bit.
- When you reduce your fader levels you logarithmically (is that a word?) reduce bit rate permanently. As a guide -40dB is down to 16-bit quality. Same goes for a 16-bit file, as you can imagine - you won't have a lot of bits left. Please note that within the DAW, unless you are bussing multiple channels out of the system (to burn a CD for example), you can move faders al you want, because the file on the disk stays the same.
- If you've taken the above in ... now perhaps you see my preference for software (plug-in) processors over hardware ones - UNLESS you use a hardware processor in the chain prior to entering a DAW environment. Just think - you record something perfectly, as close to unity as possible without clipping. In an ideal situation this would leave you at least 2dB headroom, so your file is now around 20 bit. You export this file, route it into your processor, back into your DAW. In, again, an ideal situation this file will be recorded again close to unity, leaving 2dB headroom. Now your 20-bit file is back in at 20 bit, but the audio quality is more like 17 bit, at unity. Of cause it does not sit in the mix at unity, so you pull the fader down, and when you print your mix your level is at -6dB, so now your real quality is around 15 bit.
Use a software plug-in and your quality will be maintained better.
Before some of you, once again, jump on my head with "but my Lexicon or whatever sounds much better", fine, believe that, but I can assure you that you are listening to it in isolation and not in a 40 channel mix down, when its quality will go to hell.
- Having measured the effects of summing, both by ear in a 'sound' environment as well as through measurement, I can only tell you that there seems to be an average of 6 channels you can sum (send through a bus) with minimum effect. So if you do a mix and burn a CD from 6 channels or less, the negative effect of fixed point processing will be minimal. Over 6 tracks through one bus, and the effects start to increase. 30 tracks or more and you will incur some serious harmonic problems. Well, problems, it just "doesn't sound the way it did in the mix".
- If you think (like many do) ok, now I'm going to record just my lead vocal close to unity and everything else at lower levels to fit around it, so I don't have to move my faders and reduce my bit rate / quality. WRONG. Best to start off with a high bit rate file - period. This is the only way you'll have some room for playing around.
- Clipping. Your #1 enemy. Go too hot to try and get to unity - clip - and you have a file with distortion, close to 24 or 16 bit. Now reduce the level as much as you like, you will still have a file with distortion, at a lower bit rate. And guess what? The lower the bit rate, the less your frequency range, the more your midrange comes peaking through, the more you'll hear your distortion.
- Solutions. Well, the real solution is to buy a system with high bit rate floating point processing. That's like telling someone who's got a Mackie to go and buy a Neve, so the answer is to make the best out of what you have got, and that does by no means mean bad sound (after all, consider I'm a Pro Tools man - by choice). Its like sex, safe practice works!
* We've dealt with the 'record at unity but don't clip' issue. That is a first and foremost good practice.
* We've dealt with the reasons to be careful when using outboard gear as well. Don't worry about it if your track count is low (worry about time-adjustment instead); otherwise, be careful, it has disadvantages.
* If you have a relatively low track count to mix down, don't worry a lot about summing problems. If your track count is in multiples of 6, start worrying about it IF your objective is to maintain high quality sound.
* Grouping tracks can be a solution. THE EFFECTIVENESS DEPENDS ON YOUR SYSTEM. Printed that bold, you have to try that with your own system. Do it different ways and listen to the difference. Let me give you an example. You have recorded a 5 part backing vocal harmony, which would increase your track count by 5. You can reduce that to one stereo or 2 mono files. So, you mix ONLY that group of vocals, close to unity. Note the 'only' in bold - mute everything else, all that matters is the right balance and levels for this particular group. You can bounce it and import in back into your session, all within your DAW, this brings down your track count, retains the value and quality of your vocal group and minimizes the negative summing effects later on. When you have done that you sit this group in the mix at its desired level.
Naturally you can make other groups and do the same thing. Time consuming, but it works.
A note to the above - disable the tracks you have just grouped and use your new stereo or 2 x mono group in the mix, but don't throw the old files away, you might find you need an adjustment within a group at some point late in a mix down.
* Next one, summing outside the DAW - analogue.
This is really only an option if you have good quality analogue equipment available. You could, for instance, run 6 channels into an analogue console and run the L/R bus back into an A/D converter, back into your DAW as a stereo or 2 x mono file.
Another alternative - if you have a whole bunch of good D/A converters and a very good analogue console - do all the processing you like within your DAW and do your final mix on your console.
This is one of the things a lot of high end pros do - they just use Pro Tools as a recorder and processor, and a high end console either both as a front end and mixing surface, or just the latter at final mix down. In many studios this method is now becoming a preferred option, instead of recording on multiple 2" machines, which bring with it their (large) limitation on editing and processing
* Summing outside the DAW - digital.
Plenty of ways to do this. Cheapest option, no analogue artifacts, and it works.
Personally I use an SRC192 to route my stuff through. Say we're back to the 6 channels of vocals - I route them 2 channels AES out of Pro Tools, into the SRC192, back into Pro Tools as a stereo file. No loss in sound quality.
Another way. If you have a quality piece of outboard gear, which processed at the bit rate and speed of your session - try routing your group through that and see what it sounds like. If you have a Masterlink for instance - run in at 24/48 and out into your DAW at 24/48.
If you have good converters you can even try doing a D/A and A/D chain. In some cases, especially in the case of D/A converters with high quality analogue noise shaping, like the Lucid D/A9624, this might even improve your audio quality when returned to your DAW, as it will 'smooth out' the harshness of low bit rate digital sound.
In conclusion:
1. Start worrying about summing as a result of fixed point processing if you regularly mix sessions consisting of a lot of tracks. Way over 30? Worry seriously..... If it is your objective to retain high quality sound.
2. Practice 'safe recording' - develop all the skills required to get your sound in right in the first place.
3. Be patient, summing in a way that reduces the negative effects of fixed point summing takes time.
4. If you have the patience and build the skills, there are so many things you can do to avoid this problem that it still makes DAW recording the most viable option in my opinion.
5. Getting your core system to work right is still #2 priority. #1 is described in2, above.
6. Good converters and accurate clocking will still have the most positive effect on you digital audio quality.
Hope that helped a bit.
yeeeeeeeeeee!!!!! someone understands....my fingers are not sore for nothing!