in between samples?

J

jho1986

New member
I saw someone say that there were hidden peaks between samples and they were basing that off the fact that the levels were recorded too hot...

I just want to say that there is no such thing as in between samples. If a peak wasn't captured because it was in between samples, then the only way to fix that is to record with more samples. Also, the only problem with recording too hot, assuming no clipping, is that the preamps may be working too hard causing distortion. I may be not be completely correct with what I'm saying and maybe someone could correct me if i am wrong.

BTW, I wouldn't normally start a new thread on this, but it was a closed thread that I couldn't respond in which is completely ridiculous, and I just couldn't help but to let people know who may have read that thread because a lot of what was being said was just wrong.
 
What thread did you read that in?

Anyone who thinks there is something "inbetween" samples needs to go back to digital audio 101. ;)
 
bblackwood said:
Are you suggesting inter-sample peaking doesn't exist?
Click to expand...

I was saying there is no audio data stored between samples in a WAV file. At least not from my experience when coding with standard PCM WAV files.
 
masteringhouse said:
This should help clear up some of the confusion:

http://www.cadenzarecording.com/papers/Digitaldistortion.pdf
Click to expand...

That's a good article. I never got too deep in working with raw WAV files, and never did any kind of special filters for wav reconstruction or interpolation. I guess that is done in real-time by the software or the D/A converters...?

Because the little bit of work I did with WAV files, there was absolutely no audio data between samples stored in the file.
 
danny.guitar said:
Because the little bit of work I did with WAV files, there was absolutely no audio data between samples stored in the file.
Click to expand...
Of course there's no data 'stored' between the samples, but that doesn't mean there isn't audio post-reconstruction filtering...

We don't listen to the digital signal, we listen to the analog signal post-DAC, so inter-sample peaking, while not a big issue, does exist.
 
bblackwood said:
Of course there's no data 'stored' between the samples, but that doesn't mean there isn't audio post-reconstruction filtering...

We don't listen to the digital signal, we listen to the analog signal post-DAC, so inter-sample peaking, while not a big issue, does exist.
Click to expand...

Yep, that's why anyone who tells you that you can hear "staircase steps" in digital audio will also likely be able to give you a good story about how they were once abducted by aliens.
 
masteringhouse said:
Yep, that's why anyone who tells you that you can hear "staircase steps" in digital audio will also likely be able to give you a good story about how they were once abducted by aliens.
Click to expand...


That or they are listineing to that retarted Cher song with the chromatic auto tune:D

I know that's not what you're referring to, but I could not help it.

Nice link masteringhouse;)

Edit: VERY nice link!

F.S.
 
Last edited:
from my understanding, if it peaks in between the sample, then it does not get captured in the digital audio files. Even though intersample peaks exist, it is only in the analog domain. Also, even though your digital meters may not be reflecting upon the true peak of the analog wave-form, they are only working with the converted waveforms.

All being said, I can't see any reason to even care about inter-sample peaks because once the analog waveform gets converted, they no longer exist.
 
jho1986 said:
from my understanding, if it peaks in between the sample, then it does not get captured in the digital audio files. Even though intersample peaks exist, it is only in the analog domain. Also, even though your digital meters may not be reflecting upon the true peak of the analog wave-form, they are only working with the converted waveforms.

All being said, I can't see any reason to even care about inter-sample peaks because once the analog waveform gets converted, they no longer exist.
Click to expand...


one word.....................RECONSTRUCTION!!!! the descrete steps have to be converted back to a continuous analog signal, you know , the D/A part. that's where it goes down my man. In the cd player itself.

Distortion to the people!!!!!!!!!!! :p :p :p


:D :D
 
jho1986 said:
All being said, I can't see any reason to even care about inter-sample peaks because once the analog waveform gets converted, they no longer exist.
Click to expand...

If your brain has an AES, ADAT, or SPDIF connection that may be true, otherwise they exist. :)
 
jho1986 said:
All being said, I can't see any reason to even care about inter-sample peaks because once the analog waveform gets converted, they no longer exist.
Click to expand...


Wich is milliseconds before it hits the consumers ears miles away from the studio at which point it may or may not have clipping, Unless of course you mix to tape. Oh waite it will get mastered back to digital anyway. Don't see many tape racks at the store anymore.

Ive stuck some commercial cd's in certain players that where god awfull. I now understand why some players handle them better than others and how some one could let that out of a mastering studio.

F.S.
 
Question for Mastersteringhouse:

I need you to correct or confirm a assumtion I made that was not covered in the article.

I assume that when using peak compression like a limiter you are changing the top of the wave form to where a wave form that was climbing and decending at a pretty steady rate (like a ball thrown in an arc) is changed to a wave form that climbs at a steady rate until it hits the compression and then it has a drasticly lower rate of climb and fall as it moves threw time until it goes below the compression at which point it decends at a rate similar to the rate it previously climbed at. This giving the wave form a less pointed more square (but rounded) peak.

So I again am assuming one of the reasons compression (beyond the obvious higher over all levels and more chances to clip) is so much of a culpret in the inter sample clipping/peaking is that, the d/a converters assume a steady or natural rate of climb and fall between the samples. This making a compressed wave a prime target for misinturpitation by them?

Oh, and yes I am The run on sentance King!

Hope that made sense to ya.

thanks
F.S.
 
Freudian Slip said:
Question for Mastersteringhouse:

So I again am assuming one of the reasons compression (beyond the obvious higher over all levels and more chances to clip) is so much of a culpret in the inter sample clipping/peaking is that, the d/a converters assume a steady or natural rate of climb and fall between the samples. This making a compressed wave a prime target for misinturpitation by them?
Click to expand...

Well I wouldn't really call it an assumption so much as a distortion of how they were designed to work. Let's take a simple example. A sine wave is digitized A/D in a way that the peak of the wave exceeds 0 dBFS. The result is that the wave is represented internally as a square wave instead of a sine wave. A square wave has overtones that were not part of the original sine wave, during reconstruction a low pass filter is applied that removes overtones above Nyquist (frequencies beyond what the sample rate is allowed to represent). This is done to prevent alias frequencies from being part of the output signal. The overtones below Nyquist would still be part of the output signal however. When some of these overtones are removed due to the low pass filter the peak level of the signal rises and the analog level is higher than expected. This is a basic explanation of what causes it, but it's highly dependent on the D/A converter and how it was designed.
 
Inter-sample peaking happens all the time, it's only an issue when you're flirting with 0dBFS. I mean, what are the chances that the sampling process would actually hit the absolute peak of every signal every time? Not very good...
 
masteringhouse said:
Well I wouldn't really call it an assumption so much as a distortion of how they were designed to work. Let's take a simple example. A sine wave is digitized A/D in a way that the peak of the wave exceeds 0 dBFS. The result is that the wave is represented internally as a square wave instead of a sine wave. A square wave has overtones that were not part of the original sine wave, during reconstruction a low pass filter is applied that removes overtones above Nyquist (frequencies beyond what the sample rate is allowed to represent). This is done to prevent alias frequencies from being part of the output signal. The overtones below Nyquist would still be part of the output signal however. When some of these overtones are removed due to the low pass filter the peak level of the signal rises and the analog level is higher than expected. This is a basic explanation of what causes it, but it's highly dependent on the D/A converter and how it was designed.
Click to expand...


I don't think I can explain what I mean, but I do know I was making the mistake of thinking of editing digital being like analog when I was picturing the wave forms being compressed. So You answered my question by making me think. The wave forms don't exist as a anolog form when I'm editing sooo my question is mute.

Thanks

F.S.
 
You must log in or register to reply here.
Back
Top