I shoudn't have tried 96 Khz !

nprime · Post by **nprime** » Sun Oct 03, 2004 1:37 pm

On 2004-10-03 13:58, Immanuel wrote:
I don't know *hit about, what is actually important here,

...it really is a good article that blazeboylan posted, you should read it! Very informative, if not perhaps a little deep mathmatically speaking.

We usually do not listen much to sinewaves at this frequecy, but I mypothesise, that if we are able to hear the difference between a saw tooth and a sine wave at 15kHz, then we will benefit from higher sample rates, simply because the complex sound of music will be less chunky.

...the thing is, if the fundamental tone is indeed 15 KHz, the harmonics will be out of your range of hearing, out to way beyond 20KHz, which your speakers do not reproduce. Since we require the harmonic information to determine the difference between a sine wave (no harmonics) sawtooth (mostly even harmonics) and a square wave (mostly odd harmonics), the 15 KHz tones will sound the same.

cleanbluesky · Post by **cleanbluesky** » Sun Oct 03, 2004 2:59 pm

The PDF article that was posted, while researched fairly well, makes some vast errors.

For example... "Research shows that musical instruments may produce energy
above 20 KHz, but there is little sound energy at above 40KHz..." This is someone who has done too little research into this area, if he had he would realise that different musical instruments have different harmonic distributions and while some instruments total spread of harmonics lay under 60khz, an instrument such as a cymbal has a massive spread of harmonics which are still strong even up to and past 100khz.

Also, I have a psychlogy degree and have done research on psychoacoustics and have found that there is a lot of information that is percieved in music that is not immediately describable... the idea that a human being is able to hear sounds betwen 20hz and 20khz only is based on research caried out by Fletcher and Munson, in the 1920s when they administered a simple hearing test to thousands of people and declared human hearing to havce the 20hz to 20khz range. Understanding has moved on - the simple hearing range of a human being MAY be 20hz to 20khz but there is good evidence to believe that the percieved hearing range can be much higher, and that while a person would not be able to describe an isolated frequency in the 20khz plus range, they may be able to hear the effects of intermodulation and percieved qualities of tone with harmonics beyond 20khz... what that means is that while you would peak at a hearing test at 20khz... if you were to hear an instrument with your ear it may sound different to you against an instrument that had been reproduced and put through a 20khz low pass filter.
Also, whlie Nyquist frequency is the utmost frequency that can be recorded at a sampling rate, it is possible that tones that are just below the nyquist frequncy can be distorted due to the ragged shape their wave becomes as the nyquist frequency approaches their frequency (think about jitter between a digital input and an output, if you have a note of 19.5khz and a nyquist frequency of 20khz, the points that are sampled will be slightly different due to the different frequencies e.g. if the wave starts sampling the analog singal at its crest and trough the peak/trough will have a slightly different frequency to the sampling rate so will go in and out of phase with it, giving a good representation of the wave at some points and a almost flat representation at others - a good way of testing this for yourself is to generate a sine wave and then decrease the sampling frequency in slow increments. As the sample rate decreases the note becomes distorted, then volume will pulse in and out as the the sine wave freq and the sampling rate go in and out of phase)

At 44.1khz the Nyquist frequency would be 22.05khz, which is roughly 2khz above human hearing... the musical difference being just over one tone. That would mean that a CDs frequency range is so tight to your own hearing range, that it could not record a tone and a half more than what you could hear... is that not too close? Is it not possible that someone's hearing could strech a tone, given the diversity of human life?

Also, I believe the main argument that 44.1khz is 'enough' is a little redundant. Isn't it best to make the best representation of audio as possible? Did anyone ever tell Michelangelo to use large paint brushes because that level of detail was 'enough'? Did anyone suggest that Beethoven compose on a 60-note piano because it was 'enough'?

cleanbluesky · Post by **cleanbluesky** » Sun Oct 03, 2004 3:09 pm

On 2004-10-03 06:25, ernest@303.nu wrote:
the human brain is extremely sensitive to phase information.
[ This Message was edited by: ernest@303.nu on 2004-10-03 06:27 ]

A brilliant argument for high sample rates! I agree totally, 'science' doesn't understand that much about psychological audio, and chooses to ignore a lot more...
Long live high sample rates!

Immanuel · Post by **Immanuel** » Sun Oct 03, 2004 4:03 pm

nprime

I am not talking about fundamentals. Overtones will still make the waveform "shake" at the overtone's frequency. And that "shaking" will be less acurate, if there are not very many samples.

astroman · Post by **astroman** » Sun Oct 03, 2004 4:45 pm

On 2004-10-03 16:09, cleanbluesky wrote:

A brilliant argument for high sample rates!

I'd rather call it a brilliant argument for precise clock rates

your assumption is only correct if the clock quality can keep up accordingly with the sample rate - afaik that's a technically very demanding process.
What's more significant: jitter artifacts by a shifting timebase or sampling artifacts due to a low rate ?
Both cause waveform distortions and as a consequence additional (unwanted) harmonics.

cheers, Tom

Post by **garyb** » Sun Oct 03, 2004 4:56 pm

there's no place to play this stuff back, except YOUR bedroom. the machines handle 24 tracks of this stuff poorly. end of controversy.

astroman · Post by **astroman** » Sun Oct 03, 2004 5:34 pm

On 2004-10-03 10:38, ernest@303.nu wrote:
I hope you're kidding me, or are you really that misinformed?
Super Audio CD can accomodate for 4 channels of audio (or even 6 channels, according to the specifications), but it's main feature is the so-called 'Direct Stream Digital'-technology, allowing for much higher resolution and stereo coherence (2 channels!) than the traditional 44.1kHz recordings.
This 'Direct Stream Digital' differs somewhat from 'traditional' quantized sampling techniques, but both 96kHz/192kHz recordings and SACD-recordings benefit from higher phase-accuracy, which allows for a more accurate stereo image than possible with a 44.1kHz recording.

I'm afraid you got me there - in fact I extrapolated somewhat from the '4 channel fact', assuming no one would seriously consider the release of a (new) perfect classical recording leaving that feature out.

And yes, I'm not very good on specs details - but when I read about the direct stream thing I couldn't help but think that it's a real smart marketing move.
In it's fundamentals it seemed not that revolutionary different as Sony likes put it.

Shortly before I had read a book about 'classic' digital audio technology (just for curiosity's sake) and I admit I was rather impressed about the technical complexity (mechnical, error correction, filtering algorithms, data formats etc.)

Nevertheless I suspect that a lot of the final quality of a SACD is derived from the top quality gear they usually apply in those cases - in both the analog and digital domain.

But I wouldn't consider myself as a defender of low sampling rates - I'm just an opponent of believe-in-numbers (and specs alone).

It's OK to record in 96k, but it's not OK to insist that a company has to shift it's b*tt to make THAT the most smooth running thing on earth.
At least not if the effort is ridiculous compared to the result.

Anyone is free to buy a 6 figure Sony if he or she believes that it will push their respective records

A 90dB dynamic range is a joke when everything is compressed like hell or a club vinyl player can reproduce 70db at best.

I totally agree on fine classical stuff (I have a H. Hahn record from 2002 produced with Sony DSD), but what really made me think was a copy of the Boulez version of the Wagner 'Ring' dating from the mid 80s.
I'd really like to know how this was recorded (it's live and has a tremendous quality) and they (obviously) even used it for a DVD release.

cheers, Tom

astroman · Post by **astroman** » Sun Oct 03, 2004 5:43 pm

On 2004-10-03 13:58, Immanuel wrote:
...but in these discussions my mind always go to the 15kHz sine wave topic. At 44,1kHz sampling, you have ~3 samples to create a sine wave. At 96kHz sampling, you have more than 6 samples for the same task. ...

apparently even 8 khz is enough - there's a really good Theremin (it's sine, isn't it?)fake on an 'Air' record (Kelly watch the Stars) which originates from a Casio SK1 toy sampler...

cheers, Tom

[ This Message was edited by: astroman on 2004-10-03 18:45 ]

nprime · Post by **nprime** » Sun Oct 03, 2004 6:39 pm

...does anyone know of a software frequency counter? I'd like to compare sine/sawtooth/square waves at frequencies above 15 KHz.

Fletcher....Munson...Nyquist

I'd like to do some tests of my own...

R

LHong · Post by **LHong** » Sun Oct 03, 2004 10:50 pm

On 2004-10-03 19:39, nprime wrote:
...does anyone know of a software frequency counter? I'd like to compare sine/sawtooth/square waves at frequencies above 15 KHz.

Fletcher....Munson...Nyquist

I'd like to do some tests of my own...

R

This has the built-in Frequency Counter that you might be interest? But, sorry only up to 22.050 Khz! Does it help?
http://www.planetz.com/forums/viewtopic ... orum=13&28

valis · Post by **valis** » Sun Oct 03, 2004 11:21 pm

I'm not sure that last sentence makes total sense...the difference between a 15khz sin wav and a saw wav is that the sin wav would be 15khz ONLY and the saw would have all the harmonics. In the case of the sinwave a properly designed d/a circuit should be able to reproduce the sin wav with minimum distortion regardless of whether its 44.1 or 96khz...since you're dealing with the resultant vector based off the sampling 'steps'.

There are certain caveats for filter design and the phase-ripple they impart in the upper octave but as has been stated that's a question of engineering...and imo. simply moving the passband higher is a bit of a cop-out so that your filter design can be 'cheaper', and anyone that looks for such shortcuts is likely to also use a cheaper clock too.

So in my opinion the main reason there's a push towards 96khz and higher (on the industry side) is that it reduces the cost of the filters and clock. Basically instead of improving the design/engineering/programming its simply brute-force 'upsampling'.

cleanbluesky · Post by **cleanbluesky** » Mon Oct 04, 2004 5:54 am

On 2004-10-03 17:45, astroman wrote:

I'd rather call it a brilliant argument for precise clock rates

cheers, Tom

It's a good argument for high-samplerate technology all round.

cleanbluesky · Post by **cleanbluesky** » Mon Oct 04, 2004 5:56 am

On 2004-10-03 17:56, garyb wrote:
there's no place to play this stuff back, except YOUR bedroom. the machines handle 24 tracks of this stuff poorly. end of controversy.

Hmmm. Anyone with acess to a DVD-audio player can get high-sample rate music. It's pretty easy to custom build a setup that could handle many tracks of high-samplerate (depending on budget...)

Post by **garyb** » Mon Oct 04, 2004 11:51 am

well let's see who WON'T be listening to your music.

people with cars, portable stereos, audiences with djs....

still, there are a lot of dvd players out there, true...

as to the system that can be built, the high rate-worthy set up just jumped in price about 3-5 times and complexity by 10, too rich for my blood or patience. i still say better programming and better clocks makes more sense. this is analogous to natural movements in film. it has been determined how many frames per second it takes to get good looking movies. more frames per second could have been used and film may have even looked better, but technology and $$ were factors that said, "leave it here, that's good enough". movies have not suffered.

i sound like i'm a hater and just against, but if this becomes the new standard, i'll comply....personally, i just wish that they'd spend more time getting the most out of what i already have. there comes a point when the continuing aquistition of new stuff does not balance out the lack of product output.....

cleanbluesky · Post by **cleanbluesky** » Mon Oct 04, 2004 12:00 pm

On 2004-10-04 12:51, garyb wrote:

....personally, i just wish that they'd spend more time getting the most out of what i already have. there comes a point when the continuing aquistition of new stuff does not balance out the lack of product output.....

I agree, Avalon preamps going into a ProTools rig do not make up for being a 5h1t musician...

High sample rates on everything would be a blissful dream...

nprime · Post by **nprime** » Mon Oct 04, 2004 1:02 pm

On 2004-10-04 00:21, valis wrote:
I'm not sure that last sentence makes total sense...the difference between a 15khz sin wav and a saw wav is that the sin wav would be 15khz ONLY and the saw would have all the harmonics.

My contention is that all the harmonics would be above the average adult human's hearing range. If the harmonics are inaudible you will not be able to tell the difference between the sine wave and the sawtooth.

R

[ This Message was edited by: nprime on 2004-10-04 14:03 ]

dehuszar · Post by **dehuszar** » Mon Oct 04, 2004 2:59 pm

These conversations keep popping up every 6-8 months, and they're always interesting...

My little contribution is to remind people that the ear is only a technical reference in creating and enjoying music. The most overlooked aspect of music and rhythm is that the frequencies actually hit your body and resonate through it.

So even though there are frequencies that our ears are not able to process, we can still notice them being gone. A violin is the easiest example for this sort of thing. If played right, it produces a major emotional response even though many of it's active freq's are right on the margin and above what humans can hear.

The main thrust for higher frequencies and bitrates is not necessarily just to hear more, but also to get the full path of the wave represented because we can feel, or sense the difference.

One of the main reasons most people swear by records over CDs is that the analog process captures the entire waveform, even if the phono signal gets a little noisy. CD's operating at 16/44.1kHz are almost like listening to a high-speed audio flipbook in comparison.

The difference is at 44.1kHz, the slices that are missing go by so fast that you can't point out where it is absent, but if you A/B the two together you can clearly tell which is which, even if you couldn't explain why.

I guess it could be compared to the stop motion photography that was used in the Adventures of Tom Thumb. A great watch for those who haven't seen it.

People are always complaining that audio has advanced at a much slower rate than video, but I think it's important to remember that even though our brains are primarily geared to interpret "reality" visually, we see things at a comparitively low resolution.

Even so, you can watch video that has been rendered at resolutions far higher than our eyes can perceive, and mirror movements performed by real people, but it's still incredibly easy to point out what is computer rendered versus what is real footage (though, admittedly, they're starting to get REAL close).

Food 4 thunk.

sAm

decimator · Post by **decimator** » Mon Oct 04, 2004 4:13 pm

After further tests ... here are the plans : record everything at 96 Khz ( 24 or 32 bits ? to be tested later ) then downsample at 44.1 with an anti aliasing filter then mix.

Because I noticed very often background noise over 22050 Hz so even if there's not clear aliasing, mud is added when the spectrum folds ( recording at 44.1 instead of 96 ).

By downsampling to 44.1, you also wipe off the extra dB : extreme case a " mean " level of 3 dB at 96 Khz went down to 6 dB at 44.1 because half of the " matter " was above 22050 !!

Even more extreme case ( but rarely seen ) most content above 22050 so when you record at 96 Khz : not that much to hear, at 44.1 sounds ! creative aliasing eh ?

By downsampling 96 to 44.1, I hear nearly exactly what I hear at 96 and I have less " ghost " dB.

decimator · Post by **decimator** » Mon Oct 04, 2004 6:03 pm

Grrr ! I just fell on a long streak of cases where downsampling increased the dB level, it was too easy ...

ChampionSound · Post by **ChampionSound** » Tue Oct 05, 2004 7:04 am

And what about working at 48 kHz instead of 44.1 kHz? Is that a good compromise for saving DSP and slightly improve soundquality?
Or do you get converting problems with downsampling at the final stage? I really have no clue, sorry