| Author
|
higher Khz recording
|
koalakube
IsraTrance Junior Member
Started Topics :
48
Posts :
437
Posted : Apr 24, 2006 08:46
|
Quote:
|
On 2006-04-23 21:12, texmex wrote:
for the mixdown that should be mastered, 24 or 32 bits gives you enough headroom to lower the signal so that no clipping occurs and you still have the resolution for the final master (16 bits for cd etc).
|
|
I will try to make it as simple as i can:
Stating what you have just done show a basic confusion beetwen headroom (space beetwen STandard Operating Level=+4dBu=1.23V for pro equipment and clipping point=usuaually+26dBu) and peak point.
a full loudness at 16bit"sounds the same" as a full loudness at 24bit
If your worries is the Signal to Noise ratio, people like you mentioning 24bit should srat off reading theory of noiseshaping and dithering,according to which the dynamic range of a cd can range above the 100db
Quote:
|
and you still have the resolution for the final master (16 bits for cd etc).
|
|
From 24 bit you gotta go down to 16bit anyway,and the process (at least for a purist )could be painful ;-)
There are about a dozens of different alghorythms for the down-dithering.
At this point Id like to hear from you which algorythm would you choose and why (if you are knowledgable of the difference beetwen them).
And once more, 8 bit of resolution beetwen 16bit and 24 bit are not "up" but "down".
0dB full scale (= max level for digital support) will be rappresented by a number that equal a number =14milions quantae (2^24 for a 24bit) instead of 65536 (2^16 for 16bit)) but the voltage inside a converter is the same,it depends from the clipping point (around +26dBu).
24bit gives you more resolution near 0dBfs therefore from a probabilistic point of view there is less chance of a digital error when we work near the full resolution,and eventually it gives you a more stereo and accurate stereo image (from an "audiophile" point of view)
|
|
|
Elad
Tsabeat/Sattel Battle
Started Topics :
158
Posts :
5306
Posted : Apr 24, 2006 13:20
|
|
TopDown
Started Topics :
7
Posts :
62
Posted : Apr 26, 2006 11:56
|
Quote:
|
On 2006-04-24 13:20, tsabeat wrote:
actualy , apart from fades and delay/reverb tail , i dont think i use more the 30db range.
and after hard limiter at the mastering i think it must be even less.
what does it mean ?
|
|
It means you can safely work on 8 bit recordings !!!  |
|
|
|
generator
Started Topics :
1
Posts :
12
Posted : May 2, 2006 21:53
|
Hehe, I'm way too busy and shouldn't be responding, but its become quite a fascinating discussion. I think Undertow and I have gotten to the point where we agree to disagree about the useful maximum sampling frequency to be used. (Me, I can see 88kHz or 96 being useful. Undertow would stick to 2x highest audible frequency.)
That said, one thing I will agree with, is that 196 and 384 are certainly very likely ridiculous sampling rates - probably far beyond anything we can possibly use, and just being used for marketing. However, those sample rates *WILL* (I have no doubt) drive a whole generation of new audio software and hardware. Period.
I would like to see peoples' setups ultimately jump to 24 bit/96kHz.
1. I probably won't hear a difference, especially considering the music I make.
2. I'm not even bothering to consider moving up now, because
2a. its going to take faster computers to drive the same setups we have now at those higher rates, and we're not there in terms of computing power
2b. the soft- and hardware isn't mature and standardized enough
2c. expense
However I'm still not convinced that you don't get a better picture of the sampled waveform coming in if you go about 2x highest audible frequency. After all, at that rate you're going to see a sine wave as a triangle which is clearly not a correct representation. I don't know enough about oversampling to say whether this overcomes this limitation, and can't see on a quick think how it overcomes a high freq wave being distorted as a result of too small a sampling rate.
Anyway I mainly wanted to post because I randomly (as a result of a discussion elsewhere on the topic of aliasing) found this site which, if anyone is having troubles understanding Nyquist, etc. will be very useful. It's also a great starter tutorial on this stuff.
http://zone.ni.com/devzone/conceptd.nsf/webmain/0A17E22A0530063386256F560071787E
And if you ever find that 2 ns reference, shoot it to me - I'd love to read it. It was my understanding you only got 22ms resolution (or whatever I posted previously) with 44kHz sampling. I can't see how a sampling rate of a given frequency could distinguish two sounds which are closer together than the minimum resolution at that frequency.
Mmm sound reinforcement ...
By the way interesting test of limiters. I'm not surprised to see the L2 and L3 adding higher frequencies. They add all sorts of character, which is sometimes useful in small amounts, but only on the overall output track. I find it muddies up the track if put on individual channels (as I've seen people do).
Bypass mode testing would be very interesting. I have my suspicions about that, which is why I don't plugs in bypass mode, and also undo extraneous channel connections. But that's always been a blind guess on my part :)
I would also want to test 48kHz with downsampling to 44kHz to CD versus straight production at 44kHz. That's fairly important, and opinions differ on the issue.
Cheers,
 - cosmic babies productions (cosmicbabies.ca) |
|
|
|
texmex
Started Topics :
5
Posts :
189
Posted : May 3, 2006 01:39
|
Quote:
|
On 2006-04-24 08:46, koalakube wrote:
I will try to make it as simple as i can:
Stating what you have just done show a basic confusion beetwen headroom (space beetwen STandard Operating Level=+4dBu=1.23V for pro equipment and clipping point=usuaually+26dBu) and peak point.
|
|
I admit I used the word headroom a bit out of context here (you mean analog headroom). What I meant is the space between the signal peak and the max signal level in digital realm (which is sometimes called 'headroom')
Quote:
|
a full loudness at 16bit"sounds the same" as a full loudness at 24bit
|
|
Yes, I believe I didn't argue that.
Quote:
|
If your worries is the Signal to Noise ratio, people like you mentioning 24bit should srat off reading theory of noiseshaping and dithering,according to which the dynamic range of a cd can range above the 100db
|
|
So you say that 24+ bits are unnecessary or what? I very well know what SNR and noise shaping are, and I know that 16 bits can be used to store quality audio. But those are the very last steps of the producing audio. If you want to process your tracks further (mastering yourself or by professionals) you don't want to send it dithered and get redithered. And yet you want maximum quality.
Quote:
|
From 24 bit you gotta go down to 16bit anyway,and the process (at least for a purist )could be painful ;-)
There are about a dozens of different alghorythms for the down-dithering.
At this point Id like to hear from you which algorythm would you choose and why (if you are knowledgable of the difference beetwen them).
|
|
Personally I render in 96khz/24-bits, do mastering in soundforge and then use soundforge's internal samplerate and bit-depth converters to bring the audio to 44,1khz/48khz/16 bits. I use the highest quality setting, which in bit-depth conversion is triangular dither with highpass noise-shaping. It might not be the best available, but it's good enough for me.
Quote:
|
And once more, 8 bit of resolution beetwen 16bit and 24 bit are not "up" but "down".
0dB full scale (= max level for digital support) will be rappresented by a number that equal a number =14milions quantae (2^24 for a 24bit) instead of 65536 (2^16 for 16bit)) but the voltage inside a converter is the same,it depends from the clipping point (around +26dBu).
|
|
What gave you impression that I meant something else? I think we both agree that 16 bits just is not enough bits to be processed further.
You clearly think in the analog terms, but this is pure digital. The way the signal gets realised in the converter is irrelevant (nearly irrelevant, there is always the inter-sample clipping...) |
|
|
|
koalakube
IsraTrance Junior Member
Started Topics :
48
Posts :
437
Posted : May 5, 2006 08:57
|
Texmex,
to sum up what i think people get it wrong about higher khz (im not pointing this directly to you) = loudeness
lemme put it this way:
If you mesure your member in mm instead of cm does it make it longer?
;-)
Peace
Mb |
|
|
|
texmex
Started Topics :
5
Posts :
189
Posted : May 5, 2006 15:21
|
Quote:
|
On 2006-05-05 08:57, koalakube wrote:
Texmex, to sum up what i think people get it wrong about higher khz (im not pointing this directly to you) = loudeness
|
|
Yeah. Though I don't quite get it how more khz would make it louder. For bit-depth it is understandable to have such a misunderstanding...
Quote:
|
lemme put it this way:
If you mesure your member in mm instead of cm does it make it longer?
;-)
|
|
But beware, you can get into trouble if you don't leave enough headroom between your thing and the end of the ruler. Clipping will be painful and probably very audible! ;D
Anyway, maybe someone should write a topic on these matters... khz, bit-depth, noise, stereo-image etc.
cheers and happy weekend! |
|
|
|
|
