When to normalize?

  • Thread starter Thread starter dafduc
  • Start date Start date
The point of the upsampling was for the processing I would be doing (Acoustic Mirror, noise reduction). The noise floor was high enough to compete with softer sections (of which there were none in the movement I linked, but large protions of the concert were a capella). Check out this link for an a capella recording (but mp3 is mon with only 64k resolution).

Highest possible quality in the case or THIS concert meant staying exactly where I was. The segments of the concert ranged from 16-22 minutes, going to highest resolution would have put me somewhere under 20 minutes for a stereo recording (though I may consider mono next time).
 
Upsampling does not help. It doesn't add anything.

You can change the bitrate and that will help with processing by giving you a wider dynamic range to work with. That is the whole 16bit to 24bit stuff.

Increasing the khz won't help and it just adds artifacts.
 
I did NOT know that! I thought both mattered.

Thanks for the clarification...
 
Glad to be of help. Sampling rates are a big cause of confusion.

FWIW I would pick a higher resolution mono track over a lower resolution stereo track any day.
 
OK, so if upsampling doesn't make any difference in processing audio, why would anybody ever record at a higher sample rate (ie 96kHz) to begin with if they are eventually going to resample at 44.1 for the final product? I've heard that differences between 44.1 (or 48) and higher rates like 88.2 or 96 are audible, but I suspect you would just be wasting hard disk space unless you actually plan on listening to the final mix at 96kHz. In fact, I think it would probably be worse since there would likely be artifacts from the resampling. What do you think?
 
I thought he was saying it's 'cos my original signal was only 44.1.

So recording source at 88.2 or 96 might be better, but not if you're starting with 44.1.

Even if you do processing in between - which was the new one on me.
 
There's a big difference between recording at 96khz and upsampling a 44.1khz recording.

When you record at 96khz you are recording frequencies up to 48khz. When you record at 44.1khz you record frequencies up to 22.05khz (divide in half, see the pattern?)

So with that in mind, if you don't record those high frequencies in the first place then you can't magically add them later by upsampling. Whether there is any real benefit to recording in 96khz in the first place is still a matter of much debate and is related to your resources (storage and processing) and the quality of your recording chain (frequency bandwidth). There are also many that argue that downsampling from 96 to 44.1 is worse than recording at 44.1 originally.

CD's are on the way out. Not soon but in the near future. When DVD audio or whatever comes along takes it's place we will have 96khz or better distrobution formats. If you want the music you record now to be ready for the future than it makes sense to record at as high a quality as possible.
 
Tex, you might wanna go and read this here:

http://www.vlsi.fi/other/VS_SRC/SampleRateTutorial.html

Also, many legit companies do upsampling to the audio before the "other" DSP their box offers. TC Electronics comes to mind, as does the QMetric plugin from Steinberg (QMetric offers you the choice of "Double Sampling".

While I don't make it a practice to up sample, IF the up sampling is done with a quality SRC, I can see benefits to the later DSP applied from doing so.

Ed
 
TexRoadkill said:

When you record at 96khz you are recording frequencies up to 48khz. When you record at 44.1khz you record frequencies up to 22.05khz (divide in half, see the pattern?)
Click to expand...

Yes, I understand the nyquist theorem. However, I think what is interesting is that even though you theoretically record frequencies up to 22.05khz with a 44.1khz sampling rate, at a 96khz sampling rate, you have a much more accurate representation of that 22.05khz signal - you'd be sampling every quarter of a wavelength or so instead of every half a wavelength.

Similarly, if you upsampled to 96kHz, you would have more data to work with and hence could do better math, right? This really seems analagous to increasing the word length for processing to me. Whether its audible or not, I have no idea - it would probably depend a lot on the type of processing.

TexRoadkill said:

CD's are on the way out. Not soon but in the near future. When DVD audio or whatever comes along takes it's place we will have 96khz or better distrobution formats. If you want the music you record now to be ready for the future than it makes sense to record at as high a quality as possible.
Click to expand...

Yes, this is obviously the best reason to record at 24/96 or higher even.
 
Or you could record at 88.2kHz and burn CD's at 44.1kHz (nice even factor of 88.2kHz) until DVD technology and prices catch up.
 
You must log in or register to reply here.
Back
Top