Looper's Delight Archive Top (Search)
Date Index
Thread Index
Author Index
Looper's Delight Home
Mailing List Info

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Date Index][Thread Index][Author Index]

Re: Frequencies was Re: AW: AW: OT: new Macbook wíthOUT Firewire :(



George Ludwig wrote:
> First, let me say I do most of my recording at 44.1, and sometimes at 
>48...
> 
> In a blind test you can quickly determine that the highest frequency a 
>person can hear is around 20K. These tests are typically done by an 
>Audiologist using a sine wave. From this you might conclude that it 
>doesn't make sense to record sounds above 20k because no one can hear 
>them. The problem is, music is not a sine wave.
> 
> There is a phenomena known as "beating". It's what you hear when you're 
>tuning your guitar. I quote: "When two single-frequency tones are present 
>in the air at the same time, they will interfere with each other and 
>produce a beat frequency. The beat frequency is equal to the difference 
>between the frequencies of the two tones and if it is in the 
>mid-frequency region, the human ear will perceive it as a third tone, 
>called a "subjective tone" or "difference tone". The difference tones are 
>always present, but they can be made prominent by using two high, clear 
>tones like the notes of a flute. With two flutes you can produce a "trio 
>for two flutes". This phenomenon can also be produced with one brass 
>instrument (multiphonics). If a French horn player plays one note and 
>hums another, then the subjective tone which is the difference between 
>them can sometimes be heard clearly."
> 
> It's clear from this that a recording that is brickwalled at 20k will 
>remove/alter beat affects caused by frequencies above 20k that would 
>otherwise have been manifested in the audible frequency range. If you 
>merely compare FFT's below 20k of one recording at 44.1 and another at 
>192, your FFT's will likely look identical (taking in to account the 
>differences in performance of ADA's at different frequencies). But the 
>human ear doesn't listen to FFT's. It hears the sound in the air. And 
>it's not until the signal interacts in the air that these beat frequncies 
>appear.
> 
> It's my belief that these beat frequencies, though subtle in most 
>musical contexts and certainly not perceived as a "third tone", are 
>nonetheless important psycho-acoustic cues that aid in spacialization and 
>"presence". And I'm relaly surprised that the audio community at large 
>hasn't figured this out a long time ago.
> 
> -George

Correct, it's intermodulation, actually both the sum and difference of 
two tones.  Now it gets interesting..

In the natural/analog domain we have two tones, say 21 Khz and 24 Khz.
Their intermodulation products are 24 - 21 = 3 Khz and 24 + 21 = 45 Khz.
Let's imagine the 45 Khz is really inaudible and rule it off.

We may not hear the 21 Khz and 24 Khz tones, but will hear the 3 Khz.
Now we record it with equipment cabable of say going to 30 Khz, and get 
3 Khz, 21 Khz and 24 Khz on track; all is fine.
We play it back through gear capable of reproducing 30 Khz, to be safe.

The played back 21 Khz and 24 Khz tone now intermodulate in the acoustic 
domain, and produce intermodulations at tada! 3 Khz and 45 Khz.
Again, we disregard the 45 Khz, but ooups! the 3 Khz will sum with the 
already recorded 3 Khz to a 3 Khz tone at twize the original SPL.

Hence, we should always filter off those obviously too high frequencies, 
and only record what's really needed.

Of cause I'm making an impossible statement, because the exact same 
scenario takes place when recording say 10 Khz and 13 Khz. and clearly 
we wouldn't wan't to filter those away, would we now..

The result:  Actually, we can hardly record and reproduce anything 
without at playback introducing drone frequencies at twize their 
original levels, so.. better not record at all ;)

The only way out is to actually *be* where natural music takes place!

Ok, I scratch me geek'ish beard, light the pipe and play Sherlock..
Hmmm... no, can't figure out what's missing, but since recorded musick 
kinda sounds akin to the original, I choose to disrespect the above.
What's left is that I don't have to care about not recording those 21 
Khz and 24 Khz, because, since they do produce the audible 3 Khz, which 
by sheer magic makes it onto tracks, it does get reproduced during 
playback, and tada!  no need to record the 21+24 Khz.  All is well!


Hope you had fun, I did.. :D
I'm unemployed at the moment - too much time to think..
I'll accept any audio engineering jobs (I'll even sing) - any takers ;)



> ----- Original Message ----
> From: van Sinn <vansinn@post.cybercity.dk>
> To: Loopers-Delight@loopers-delight.com
> Sent: Thursday, October 16, 2008 11:11:38 AM
> Subject: Re: AW: AW: OT: new Macbook wíthOUT Firewire :(
> 
> Rainer Thelonius Balthasar Straschill wrote:
> 
>>>Not wanting to start a discussion about sample rate, but..
>>>Again, on rec.audio.pro, it was mostly agreed that sample rates above
>>>44.1 or 48 doesn't yield anything, other than you her the 
>>>drive more ;) I'll look for a link to that discussion..
>>
>>
>>Well, you just started it ;)
>>
>>Without wanting to go into too much detail here, I'd like to point out a 
>few
>>items regarding sample rates here:
>>
>>As you (most probably) all know, the human ear can only hear up to about
>>16kHz (depending a lot on age and possible abuse), with some people able 
>to
>>hear up to about 20kHz (and I don't want to discuss precise values here, 
>so
>>if these values are in your opinion not correct, this will not affect the
>>text about to follow...).
>>However, these values rely to stimuli with sine waves. Other experiments
>>however lead to the conclusions that:
>>    1. the human ear can hear and discern properties in transients which
>>correspond to fourier transforms of above 20kHz
>>    2. the human ear can detect phase relationships on a scale smaller
>>than 1/20kHz in the time domain
>>
>>What do we need this for? (1) is used a lot for defining the 
>characteristics
>>of the sound. (2) is vitally important in directional hearing.
>>
>>Furthermore, we got the issue with the anti-aliasing filters, which due 
>to
>>their nature are NOT brickwalls at the Nyquist frequency (in case of a CD
>>22.05kHz).
>>
>>So taking all of that into consideration, I can see (not that I actually 
>did
>>listening tests here) that it may very well be possible that there is an
>>advantage of 96kHz over 48kHz - other than loading my computer.
>>
>>    Rainer
> 
> 
> This is common belief, often debated and not holding water.
> That we under very ideal conditions might be able to detect freqs higher 
> than 20Khz may be proven true, but AFAIK has never proven by experiments 
> related to music, even not by blind tests on Golden Ear test subjects.
> I don't have links to docs derived in tightly controlled labs proving 
> claims in either direction, which is often the case in such discussions.
> However, as most even halfways adults can hardly detect freqs above ~14 
> Khz, I find it mostly irrelevant for practical music production, even 
> when using near state of art equipment.
> 
> WRT the Shannon and Nyquist criteria, stating frequencies reproducable 
> up to half the sampling rate, provided infinitely steep anti-aliasing 
> filters are used, this used to be a big problem in older days when using 
> analog  filters, but is not the case with todays 64x oversampling and 
> digital filters.
> 
> I'm sure some will say 96 (or 192Khz) sounds much better, and wil not 
> argue that.  However, it was mathematically proven on rec.audio.pro the 
> difference is hardly there. Further, certain interfaces actually does 
> sound better/cleaner/whatever at higher sample rates, which is simply 
> due to inadequate electronics desighns.
> 
> 
> Please note: I honestly didn't mean to start such a discussion, merely 
> point out that if we assume 48Khz is enough, dealing with a more than 
> decent numbers of track over firewire400 shouldn't be a problem.
> 


-- 
rgds,
van Sinn