eratu


I'm quite late to the party, but I have a fun point I want to throw in to the discussion... the original discussion of bits and bytes and frequencies, that is.... not the validity of production credits of someone. :)

There are THREE really great situations where I would DEFINITELY record at 96KHz or even 192KHz.... or even 384KHz. :)

Those reasons include for creative sound design, for scientific study, or for low-latency advantages... for example:

If I had a microphone and preamp (or device) that could capture at those ridiculously high frequencies that humans can't hear, I'd want the added resolution to capture those frequencies so I can use them for creative sound design. Basically, I could play back the 96K or 192K or higher audio file at 1/2 or 1/4 or 1/8 or 1/16 of the original rate (note, I don't mean re-sampling here, I mean slow it down the old fashioned way), and then I could bring those high frequencies into the human range.  There could be some really fascinating things up in those registers that when slowed down to our range, could be extremely useful for sound designers. :)

I've wanted to try this out just for the fun of it for years, but haven't had the time/chance/gear... wouldn't if be fun to see what goodies are in those high frequencies that I could use for sound design? :)

You can see what I mean by the potential usefulness of this by creating a sample audio file in Sound Forge, for example, at 192KHz. Then use the simple synthesis tool to create a sine waveform sweep through 96Khz. Obviously, your speakers won't be able to reproduce that (not to mention that your ears can't hear it), but then just play it back at 11KHz, and you'll hear what's there. Well, you won't hear anything special because it's just the sine wave slowed down, so it's still a sine wave. But if you are capturing something in nature at those frequencies, and then slow it down/play it back at a much lower sample rate, you'd understand what I'm talking about... that data -- those waves -- suddenly become useful.

For example, NASA often records extremely high-frequency recordings -- either audio or even way above optical wavelengths, and then shifts them into ranges that we humans can see and hear. There's all sorts of cool $h*t flying around and through us every nanosecond, and waves are waves.... just sample the waves at whatever frequency is possible to capture them, and then bring them down into the audible human range. Cool stuff ensues. Planets sing. Stars make music. I've heard various NASA recordings of such things, and it's beautiful. Put science and art together, and sound designers can have fun.

Another reason to use a high frequency like that is for latency purposes. Some people want to get better low-latency performance out of their audio device by running at a higher sample rate if the drivers can handle it at a certain buffer setting. :) But that has nothing to do with the merits of the sampling frequency for audio integrity purposes, and frankly, with the level of gear that most of us have, we're discarding most of the stuff above 20K anyway before it even hits the converters.

Having said all that, since I am human (most of the time) and I am recording human things, I almost always record at "normal" human ranges and I honestly think that 99.9999999% of the population couldn't tell the difference between 48KHz and 96KHz. There are going to be many more issues in the signal path and production process that are going to affect the audio far more than that, so I think there are other issues to tackle in the mixing path.

If you like 96K go for it. But just remember that your recording is only as good as the weakest link in your chain. Your mic, your cables, your preamp, a certain resistor in your preamp, a certain pot, a certain tube, an AD converter, the filters on that AD converter that you might not know about, your DAW mixing engine and the secret messy DSP code that some intern left in there, the summing algorithms, the DA converters, your monitors, your ears, etc., etc., etc., etc.

My two bits. Please ignore if not interested. :)

dontletmedrown


Cool.  Thanks for clarifying that Freddie.

Thanks Dave!

Muziekschuur at home

In the nineties the same was happening with consoles. It was about how many khz their preamps and mixingconsoles let through.
In the end SSL Neve and D&R (among others) created great consoles wich let through up to 100khz and sometimes up to 200khz.

If you have a lowpass filter to cut down noise in a cd-player it damages some high frequencies because of the lack of steepness of analog filtering.

44.1 was a wrongly chosen sampleformat created by Philips.

When you record music with high frequency content (every bang on cymbals will give frequencies up to 50khz (and above but much more weakened). When recording at 44.1. khz. (every sample is half a wavefile so 44.1 can hold up a frequency response of up to 22,5 khz. We loose much of of the high frequency content when loosing those 50khz to 22.1 khz content. We allso loose the lower harmonics of that part of the signal.

And by the way. We are sensitive to phase up to 200khz.

So, much of the localisation content of the signal is lost when recording at 44.1 and 48khz.

This is what the managing director of D&R (Duco de Rijke) had to say about this subject (about analog consoles).
Taken from the D&R website.... I think it is pretty interesting:
--------------------------------------------------------------------------------------------------------------
Did YOU know why analog sounds better than digital?

D&R started long ago (1972) making mixing consoles equipped with tubes.

One of the nicest parts of these hot valves were that they sounded great.

The sound was pleasantly warm and overload was not an issue with over 300 volts on the power rails and a bandwidth of up to 100kHz.

Then we experimented with transistors and could not get the same sound as we had with tubes.

Imagine listening to a sine wave of 1kHz, this should be a nice pure round sound like a pure tone of an old Hammond organ. Then switch this 1kHz frequency to a square wave, now you are listening to a sine wave plus on top of it lots of odd harmonics such as 3rd, 5th, 7th and so on .

Actually you are listening to 1 kHz plus on top of this on a very low level frequency 3khz, 5kHz , 7kHz 11khz and so on. So the difference in perceived sound is quite clear to any listener, nothing mystical here.

Now switch the fundamental 1kHz sine wave to let’s say 12kHz. Again a pure very high frequency tone could be noticed.

But now the funny part of it. If this 12kHz sine wave is switched to a square wave we all still can hear the difference between sine and square wave.

THIS IS AMAZING.

If we all agree on this, it means that we can hear or at least notice signals beyond 20kHz.

Remember the first harmonic content above 12kHz is 24kHz , and then 36kHz and so on.

These generated frequencies above 20kHz (even in small amounts) prove that it is important to have an audio system that has a bandwidth well over 20kHz.

It is absolutely a fact that a small lift in frequency response of audio equipment well above 20kHz could easily give you the impression of giving you more air or transparency.

Can you imagine why most digital consoles today sound harsh and can not even touch the smoothness of well designed analog consoles. This is easy understandable now with frequency responses that sharply fall off at 20kHz. No information above 20kHz means no warm open sound.

  Imagine what a digital upper shelving high frequency control is doing when lifting those frequencies by 16dB just before they sharply fall off at 20kHz.

I think that digital technology has along way to go before it can even touch analog audio sound wise.

Here is the thread about this subject I started a year ago;
http://www.gearslutz.com/board/high-end/128405-hearing-above-20khz-analog-consoles.html

More importantly. Bruell & Kjaer did create microphones in the nineties who could capture that high frequency content. Up to that point high frequencies could not be captured in that accuracy. (allthough tape machines record up to 50khz.

Bruell & Kjaer now has a brand called DPA microphones. When you hear those microphones (like the 4006 or the 4090) you soon realise that high frequencies add alot to a mix....

Jose7822


All this high frequency hogwash (nice word :-P) goes down the drain when you think about the plaback systems they'll be played on. 

Even the high end ones don't go much above 20KHz, so what's really the point (besides the ones already mentioned that is)?

UnderTow

Muziekschuur at home

In the nineties the same was happening with consoles. It was about how many khz their preamps and mixingconsoles let through.
In the end SSL Neve and D&R (among others) created great consoles wich let through up to 100khz and sometimes up to 200khz.

This makes sense because having your -3 dB (or whatever) point at 100Khz or more means a very flat linear response in the audible band.

If you have a lowpass filter to cut down noise in a cd-player it damages some high frequencies because of the lack of steepness of analog filtering.

Non-sense. Only someone that doesn't understand modern converters would say such a thing. The ADC is oversampled and DAC is upsampled so the analogue filter does not have to be so steep. The oversampling can be anything up to 1024 the base rate.

44.1 was a wrongly chosen sampleformat created by Philips.

You can not say that without knowing why that sample rate was chosen. It was most probably an engineering decision and as most engineering decisions, a compromise between many factors. (Btw, the Beethoven 9th or 5th or whatever story is an urban myth).

When you record music with high frequency content (every bang on cymbals will give frequencies up to 50khz (and above but much more weakened). When recording at 44.1. khz. (every sample is half a wavefile so 44.1 can hold up a frequency response of up to 22,5 khz. We loose much of of the high frequency content when loosing those 50khz to 22.1 khz content. We allso loose the lower harmonics of that part of the signal.

In a properly designed system we only lose inaudible frequencies.

And by the way. We are sensitive to phase up to 200khz.

So, much of the localisation content of the signal is lost when recording at 44.1 and 48khz.

This is absolutely NOT true. Read one of my previous posts about timing/phase accuracy in digital audio. It goes way beyond what the human ear can detect even at 44.1 Khz.

This is what the managing director of D&R (Duco de Rijke) had to say about this subject (about analog consoles).
Taken from the D&R website.... I think it is pretty interesting:
--------------------------------------------------------------------------------------------------------------
Did YOU know why analog sounds better than digital?

D&R started long ago (1972) making mixing consoles equipped with tubes.

One of the nicest parts of these hot valves were that they sounded great.

The sound was pleasantly warm and overload was not an issue with over 300 volts on the power rails and a bandwidth of up to 100kHz.

I think he means pleasantly distorted. We moved on from tubes because they were not clean enough.

Then we experimented with transistors and could not get the same sound as we had with tubes.

No indeed, good transistor designs are much cleaner.

Imagine listening to a sine wave of 1kHz, this should be a nice pure round sound like a pure tone of an old Hammond organ. Then switch this 1kHz frequency to a square wave, now you are listening to a sine wave plus on top of it lots of odd harmonics such as 3rd, 5th, 7th and so on .

Actually you are listening to 1 kHz plus on top of this on a very low level frequency 3khz, 5kHz , 7kHz 11khz and so on. So the difference in perceived sound is quite clear to any listener, nothing mystical here.

Now switch the fundamental 1kHz sine wave to let’s say 12kHz. Again a pure very high frequency tone could be noticed.

But now the funny part of it. If this 12kHz sine wave is switched to a square wave we all still can hear the difference between sine and square wave.

This points to either a bad wave generator or the signal is going into a non-linear device causing inter-modulation distortion. EDIT: Or the signals were not level matched for the in-band frequencies. We can NOT hear the harmonics. If we can hear anything, we are hearing artefact within the audible bandwidth.  Mr Neve has designed some great analogue equipment but his comments on digital audio make me wonder if he really understands how it all works. Even some of his remarks and conclusions about analogue equipment are very strange to say the least.

THIS IS AMAZING.

No. This is a flawed test.

If we all agree on this, it means that we can hear or at least notice signals beyond 20kHz.

Remember the first harmonic content above 12kHz is 24kHz , and then 36kHz and so on.

These generated frequencies above 20kHz (even in small amounts) prove that it is important to have an audio system that has a bandwidth well over 20kHz.

Again, no. This is a flawed test. Double blind tests with good wave generators have demonstrated that people can not hear the difference when you filter out stuff above 20 Khz.

It is absolutely a fact that a small lift in frequency response of audio equipment well above 20kHz could easily give you the impression of giving you more air or transparency.

Again if this happens, this is due to changes in the audible band. Probably a phase shift but possibly a saturating (thus non-linear) stage somewhere in the EQ or elsewhere in the signal path.

Can you imagine why most digital consoles today sound harsh and can not even touch the smoothness of well designed analog consoles. This is easy understandable now with frequency responses that sharply fall off at 20kHz. No information above 20kHz means no warm open sound.

This is complete non-sense. If digital consoles sound harsh (which is not true for the good ones) it has nothing to do with the limited bandwidth. On the contrary. It goes even further than that, I heard a story once by someone that was one of the first listeners of a new Neve console. He and the other listeners found it harsh and aggressive sounding. Mr Neve said he could fix that. At the next listening session everyone agreed that it sounded much better, much warmer. When asked what he had changed Mr Neve answered that he had added low-pass filters to every channel.

  Imagine what a digital upper shelving high frequency control is doing when lifting those frequencies by 16dB just before they sharply fall off at 20kHz.

It depends of the design.

I think that digital technology has along way to go before it can even touch analog audio sound wise.

Only if you don't understand it and anyway, what are the alternatives as a storage medium? Tape? Tape is MUCH more coloured than even the cheapest Soundblaster sound card.

Here is the thread about this subject I started a year ago;
http://www.gearslutz.com/board/high-end/128405-hearing-above-20khz-analog-consoles.html

Cool. Will read. :)

More importantly. Bruell & Kjaer did create microphones in the nineties who could capture that high frequency content. Up to that point high frequencies could not be captured in that accuracy. (allthough tape machines record up to 50khz.

Bruell & Kjaer now has a brand called DPA microphones. When you hear those microphones (like the 4006 or the 4090) you soon realise that high frequencies add alot to a mix....

It could be that those microphones just sound different. Or maybe they are more linear in the audible range and what you hear is LESS artefacts from ultrasonic frequencies.

UnderTow