There are look-ahead limiters, like Waves L1, etc. that can do something very similar to clipping. If overused, they create a similar distortion and waveform appearance. Then there is actual clipping, which some mastering engineers prefer the sound of. That is, they may prefer the sound of clipping to other options such as limiters, when the customer wants the CD to be as loud as possible. If the customer asks for the CD to sound great rather than loud, mastering engineers are quite happy to comply with that preference.
The -1 dB is probably a safety factor added to assure playback on older equipment. Some people have been known to use -0.5 dB for that purpose.
in the "old days" and the plant was adhering to Red Book, an excessive number of full scale digits in a row (ie clipping) would get the project bounced, I think it was 3.
So people dropped the level to -0.3db or so, and viola, no rejections.
and the plants stopped caring.
I know the disc are out there, just can't remember what. Its not the music I listen to.
Cheers
Alan
So people dropped the level to -0.3db or so, and viola, no rejections.
and the plants stopped caring.
I know the disc are out there, just can't remember what. Its not the music I listen to.
Cheers
Alan
A simple check would be a sinus of Fs/4 with sample points phase and amplitude of the samples choosen for the worst case (i.e. intersample overs). Then you will know if the plugin just monitored the digital scale or does something more elaborated.
That will be the datasheet example schematic, while the part may be used by higher fidelity units, then one can notice it wasn't using the datasheet example schematic.
Some codecs are more relevant than others. But there is some inconvenience. For instance, Foobar2000+WavePack in lossy mode, does still have large files and limited player support. But, like the NOS DAC's, the errata goes missing without missing any of the music.As for occasional ticks and pops, they have a hard time creating a "sound". BTW the over has a hard time being ON the CD it can only happen in the post processing. Maybe that explains the popularity of NOS DAC's.
It doesn't do the pre-echo or fuzz.
It is possible to nip clips at the audio amplifier, but only when it is at full blast. Indeed, it is much more likely to succeed doing that signal repair task with greater regularity at the preamp.
For succeeding at the needful aspects of audio, there's a wide variety of circuits and much sport to it.
However, it isn't quite clear as to which circuit has done what, since modifying the signal (even if to clean it of common non-audio content) is frowned upon. So, even though the capacity to replay more music than not, is always a goal, that can't be reported on directly.
Jackob: You are right. I should really to that. I don't trust the DR results, but possibly I think they use a combination of crest factor and averaging. For me if a piece of music starts out at -50dB and peaks at -0.1 then that's a DR of 49.9, not the 12-14 it usually reports. But it is a handy ready reckoner.
Did i express any worry?
It was just an example to illustrate that it was and is quite easy to get signals in the chain that shouldn´t be there.
I´m sorry that i don´t have any explicit examples at hand. My concerns were actually based on a discussion in a german forum a couple of years ago. One of the members were doing techno and typical dj stuff (everything done with software) and related to musical effects said, he could do something this way and ended up with samples that violates the boundaries of red book in the same way a one sample impuls (0000FS0000) does.
I was quite surprised as i would have thought the software developers would check for things like that.
The test signal will show if the peak value of foobar is correctly based on a resampled waveform or only uses the sample amplitude.
The DR values of these meters are usually based on psychoacoustical loudness evaluation, so are different from the traditional approach that does not reflect the perception.
There is a VST plugin component available for Foobar: Foobar2000:Components/VST adapter - Hydrogenaudio Knowledgebase
Also, this one: http://www.yohng.com/software/foobarvst.html
There are also some freeware VST plugins that can detect intersample overs. Haven't tried it, but this one might work with Foobar VST: K-Meter | Martin Zuther
Also, this one: http://www.yohng.com/software/foobarvst.html
There are also some freeware VST plugins that can detect intersample overs. Haven't tried it, but this one might work with Foobar VST: K-Meter | Martin Zuther
This would mean assuming the resampling software always does the wrong thing. If you are examining 16bit data at 44.1kHz there is nothing between the samples. I would use the term CoolEdit uses, "Possibly clipped samples".
Seems to be a misunderstanding.
Billshurv only had foobar´s peak value of -0.1dB and we do not know if that is a foobar miss or if simply no oversample overs were "buried in the data" of the cd.
If we are using a sinus with peak amplitude of -0.1dBFS (frequency of Fs/4) and sampled at pi/4 points it provides the worst case for intersample overs.
If Foobar´s peak value is still -0.1dB it uses internally a resampled representation for the calculation of the peak value, it the peak value is ~-3.x dBFS it is simply taken from the digital sample value. To be sure we should check against a 1kHz to see if Foobar is using the rms value.
As you´ve said there is no data on the cd between the samples (although there is something between the samples
) "possibly clipped samples" is a miss as the samples do never clip (in the case of intersample overs).
Yes there is a misunderstanding, I'm only considering cases where failure occurs. Modern DSP can handle the problem by accounting for the needed extra headroom. Bad digital processing of the signal before recording happens for sure, nothing much post facto can be done about it.
Hi George, in your attached picture, on the right side, there's some kind of SW showing a signals corresponding value in bits, can we know the name of the SW used and where can we find it, thanks.
I understand that you consider only cases where failures occurs; and of course nothing can be done afterwards if signals violating the boudaries got into the chain, but that wasn´t the point, as there were doubts if such signals even could/will exist.
But i was mainly surprised by your citation and the conclusion as you wrote:
"Originally Posted by Jakob2
The test signal will show if the peak value of foobar is correctly based on a resampled waveform or only uses the sample amplitude."
So my sentence gives the premise and yours the conclusion, based on the premise and i simply don´t understand why the conclusion should follow, could you elaborate a bit?
The ideas also? It always seemed to me the best way to look at it was this way. Regardless of obsolete it makes good reading. On target for this thread. Some try to say Newton obsolete after our newer scientists arrived. As NASA told me, Newton was good enough for them when Voyager.
I found this when helping a friend with zero loop feedback. Not what I was looking for, good all the same.
If someone ( not me ) wanted to do an instructive thread on feedback types that could be fun. Keep it simple as people come to learn at DIY. The complex bit later.
I found this when helping a friend with zero loop feedback. Not what I was looking for, good all the same.
If someone ( not me ) wanted to do an instructive thread on feedback types that could be fun. Keep it simple as people come to learn at DIY. The complex bit later.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.