View Single Post
Old 28th October 2010
Lives for gear
I will do short answers to these for starters, all I will say is my opinion only. As ever - longer answers are only needed where people disagree and require detailed discussion.

Originally Posted by M2E ➡️

So, I'll start it off…..

1. Is This Truth Or Myth? -
Digital can achieve and/or surpass the best analog headroom.
Truth - because the dynamic range of digital is limited only by mathematical precision, whereas analogue is limited by the physical world which cannot be avoided.

A 24bit signal path with 144dB of dynamic range is just about as good as any analogue process could practically produce.

To provide headroom simply modulate at lower levels to allow overshoots to not be clipped off in fixed point systems and output media.

2. Is This Truth Or Myth? -
Digital can truly and fully emulate analog mixing boards of yester year (SSL/NEVE etc).
Truth - if someone went to the trouble of doing so.

Providing the same result is produced for all signals it does not matter how the processing is done. There is no magic.

The biggest single difference between analogue systems of old and the current digital environment is that analogue systems had headroom above the modulation levels - to accomodate over signals and progressive analogue degradation at over normal levels.

A properly implemented digital system does not have this as it is 'correct' at all levels.

Ths changes the workflow significantly and unfortunately encourages people to modulte at maximum, thereby removing degrees of freedom from the work flow.


3. Is This Truth Or Myth? -
Digital system's hardware only achieve 20hz - 20khz, there is no reason to have a plugin @ 10hz - 40khz? We can't hear and/or feel the difference.
Truth - if everything is done correcly.

Digital hardware even at base sampling rate can provide 0Hz - 20KHz.

I personally have no evidence (however much I've tried) that anyone can hear directly beyond that range.

Therefore it seems there is nothing much to be gained from producing supersonic freqs we can't hear that serve only to cause errors in repro systems.

However being able to set a centre freq of above 20KHz in something like an EQ still has merit because the tail of the response below that can still be heard and appreciated. But this can be done without sampling the systen at higher rates.

4. Is This Truth Or Myth? -
Current metering plugins/systems are totally accurate? Which one would you recommend by the way.
Myth - unfortunately.

Current digital metering systems in workstations measure sample value only. Since this is not decoded signal the meters do not show actual signal level.

On reconstruction (in a DAC for instance) combinations of high level sample values that do not cause overs on the workstation metering can produce signal levels up to +5 or even +6dB above full modulation when decoded. How these get treated in the users equipment is not specified or precise - so errors can be generated that sound bad.

This is commonly called 'inter-sample peaking'. But really it is simply the result of increasing sample values beyond what could be generated by converting signal from the real world, and they are in error of strict sampling theory.

5. Is This Truth Or Myth? -
64bit will help the sound of older and/or current plugins.

If you mean the latest '64 bit' operating systems, the answer is no.

It refers only to the ability to access greater than 4GB of RAM and has no bearing on mathematical precision, which has not changed.

6. Is This Truth Or Myth? -
Pro Tools 48bit Mix engine is enough to handle a 192 channel mix accurately? It would not have to be a 56bit, 64bit or even a 72bit mixer.
Truth. In fact the 48bit processor has a 56bit acumulator so that accuracy is not lost when signals become large in value. There is no problem with the mixer provided it's presented with 24bit signals.

7. Is This Truth Or Myth? -
In Pro Tools, as you know, On a TDM system, plugins handle like this
TDM = 24bit in to (If plugin is double precision) 48bit to 24 bit out
RTAS = 32bit in to (If plugin is double precision) 64bit to 32bit out
TDM to RTAS to TDM = 24bit in to 48bit to 24bit out, to 32bit in to 64bit to 32bit out, to 24bit in to 48bit to 24bit out.
So it is not good to use TDM to RTAS or RTAS to TDM as too many calculations will only make sound worst and/or Pro Tools mix engine can not handle
all these calculations and will only make your mix/recording worst.
Myth - practically speaking.

The TDM buss has 24bit wide precision, which means that any signal presented to the bus has to be in that format.

Because properly designed plug-ins can process in double precision (or host 64bits for RTAS), the 24bit output can be the only source of reduction of precision and increased noise.

However a 24bit signal with dither is capable of 144dB of signal to noise ratio.

And since noise from multiple sources adds up by 3dB for every doubling of the number of sources (20Log(number of sources^0.5)) a very large number of 24bit sources are required to impact significantly on the signal to noise ratio of your mix.

For instance the signal to noise ratio of 256 24bit sources added together is

144 - 24 = 120dB

This is still around 30dB better than your output media.

So there's not much to worry about.

8. Is This Truth Or Myth? -
It is better to stay totally ITB then to use a summing mixer for width/depth/headroom. Minus color of course.
Theoretically truth. One thing a computer can do is add stuff up! Analogue struggles with it.

Also having to convert into analogue and back into digital again is obviously a big potential source of error and signal quality loss.

9. Is This Truth Or Myth? -
Even with today's cpu's, there is no way that we could exceed the sound of old reverb units. (Lexicon 480L/PCM90 etc)
Obviously this is a myth. The only thing that determines what reverb does is processing. The average home PC has many times the processing power of one of these legacy units.

The only thing preventing a plug-in being identical to these (if that is really desired) is that no one has actually bothered.

It is pretty obviously possible to exceed the performance of these units on any home PC.

10. Is This Truth Or Myth? -
64bit floating point is better than 48bit fixed point. In fact, we should go 72bit to make it more analogy.
Myth practically speaking. A 48bit fixed point system is largely equivalent to a 64bit float system in real signal accuracy over the range of -1 to +1 - because the 64bit float system has a 52bit mantissa and a 11bit exponent.

The only case where this might affect things is in recursive processes (like filters etc.) where a floating system might benefit from the exponent.

However with 288dB of real dynamic range in a 48 bit system (float or otherwise) it's not an issue at all.