JSmith
Master Contributor
Yeah... I just thought it can be handy for other readers to have links to the other past threads present.
JSmith
-
Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Are inter-sample overs the new audiophile bogeyman?
- Thread starter antcollinet
- Start date
JSmith
Sokel
Grand Contributor
- Joined
- Sep 8, 2021
- Messages
- 10,660
- Likes
- 11,557
Who's audibility? At what conditions?
And what combination to avoid?
Let's make a hypothesis.
I have a KTB here, right? And let's say I have one of those dreadful recordings (luckily I listen to classical, no worries there but still)
And let's say that this recording is PCM and not DSD so to make KTB suffer as it's low level nasty grass is both level and frequency dependent, you can make it dance at the analyzer at any fashion you like.
Yes, I know it was highly regarded with its -110db THD+N and it sold so much that someone even put it in a alu chassis and was selling it as silver.
But the truth is it's crap,
And now combine those, add the analog sins after that, say I have barely adequate amp and guarantee me bliss.
Should I believe you?
Nope.
OHtaru
Member
- Joined
- Oct 3, 2021
- Messages
- 71
- Likes
- 99
How did you confirm this? My test results show that reducing the volume can avoid ISP clipping. I tested with 11025Hz +3dB ISP at 44100Hz, there is no clipping if the volume is lower than -3dB.
KSTR
Major Contributor
I can't find it at the moment but I once posted a track here for A/B testing of IS-overs being faithfully reproduced vs clipped. With that knowledge and thorough ABX testing I remember only one person could detect the difference.
My conclusion is that with DACs that handle IS-overs gracefully by simply (soft-)clipping them as most of today's DACs do, IS-overs are non-issue, practically. A few DACs are not well-behaved and the moment digital pre-processing of any kind is involved, starting with simple EQ, we are subject to the competence of the programmers to avoid nasty IS-over effects happening upstream of the DAC.
Resampling is one of the typical offenders, and sadly even otherwise excellent software suffers from problems here (the Linux SoX utility/library for example, where they made the fatal decision to use integer math only, internally).
My conclusion is that with DACs that handle IS-overs gracefully by simply (soft-)clipping them as most of today's DACs do, IS-overs are non-issue, practically. A few DACs are not well-behaved and the moment digital pre-processing of any kind is involved, starting with simple EQ, we are subject to the competence of the programmers to avoid nasty IS-over effects happening upstream of the DAC.
Resampling is one of the typical offenders, and sadly even otherwise excellent software suffers from problems here (the Linux SoX utility/library for example, where they made the fatal decision to use integer math only, internally).
solderdude
Grand Contributor
I suppose limiters used in the production process are a bigger audible issue than the ISO (when not handled correctly).
I suppose it also depends on how 'big' the ISO over is, the duration and how often it occurs.
When that is often and substantial the recorded music is poorly produced (loudness wars and clipping) that it would be more of a source sound quality issue that is audible than the actual IS over ?
I suppose it also depends on how 'big' the ISO over is, the duration and how often it occurs.
When that is often and substantial the recorded music is poorly produced (loudness wars and clipping) that it would be more of a source sound quality issue that is audible than the actual IS over ?
Sokel
Grand Contributor
- Joined
- Sep 8, 2021
- Messages
- 10,660
- Likes
- 11,557
Lets say we have both.
How does this works?
Do they add, subtract, multiply or else?
How it would look like at signal level?
solderdude
Grand Contributor
clipped.
JSmith
Master Contributor
My NAD M33 has this issue when certain settings are on and a subwoofer is used. It requires playing specific songs that trigger intersample overs. If I’m not looking for the distortion, it’s hard to notice it.
Sokel
Grand Contributor
- Joined
- Sep 8, 2021
- Messages
- 10,660
- Likes
- 11,557
Agree, what I'm asking is how would clip² works, the mechanism behind it.
solderdude
Grand Contributor
When a sample is clipped in the recording (results in an ISO) a DAC that has headroom will create a signal that is closer to 'de-clipped' signal that has headroom left.
The thing is though that when a producer/mixer decides to 'clip' (and usually also during mixing compresses the dynamics to make the sound 'louder' and 'more dynamic sounding by actually reducing dynamics) the waveform (near its peaks) is already 'compromised' technically.
It depends on the compression being used and the kind of clipping that is used and how loud it is mastered.
When it happens a LOT and more severe it is much more 'sound quality degrading' than when it is an occasional hardclip.
Hard clipping bass notes are also much more audible than an occasional signal peak (afterall all signals are 'added') of a coinciding peak of a bass note (largest amplitude) with mids and treble shortly peaking all together and only 1 or 2 samples 'wide'.
The thing is though that when a producer/mixer decides to 'clip' (and usually also during mixing compresses the dynamics to make the sound 'louder' and 'more dynamic sounding by actually reducing dynamics) the waveform (near its peaks) is already 'compromised' technically.
It depends on the compression being used and the kind of clipping that is used and how loud it is mastered.
When it happens a LOT and more severe it is much more 'sound quality degrading' than when it is an occasional hardclip.
Hard clipping bass notes are also much more audible than an occasional signal peak (afterall all signals are 'added') of a coinciding peak of a bass note (largest amplitude) with mids and treble shortly peaking all together and only 1 or 2 samples 'wide'.
KSTR
Major Contributor
Point taken. IME IS-overs typically happen with tracks that are already "broken by design" and the perceived advantage of those IS-overs being faithfully reproduced or not is nil, probably even with extreme cases.
The peak-limiters used today are very fast and use a few samples of look-ahead time which makes them behave very similar to simple clipping, just that some scaled-down higher-frequency remains of the original waveform is riding on top of the "capped roofs".
Most of the loudness war comes from multiple cycles of upward compression of the medium and low level segments of a track, often with multiband compressors. The capping of the peaks just maximizes overall level but has little impact on the perceived loudness and dynamics.
Example songs would be nice .
I think you both are onto something , it can possibly be an actual issue with digital processing involved , with todays more complex products with EQ and DSP of all kind .
Thanks for mentioning an example product did not NAD had some buggy software in the past .
Soo maybe some test procedure should be used by Amir when testing products that do more than simply DA ? there seems to be opportunity to fudge things badly by software .
morillon
Major Contributor
- Joined
- Apr 19, 2022
- Messages
- 2,104
- Likes
- 510
As for audibility, it's essentially linked to the nature of the music and the level of expectation, etc., so it's rather "cultural"...
Engineers in rock, pop, etc., have understood this and disregard it...
In "classical" or demanding music, etc., they pay attention to it, especially since it requires precautions given the potentially significant dynamic ranges...
The constant "rattle" about audibility... in the so-called "general population" is just excessively demagogic ;-)
Engineers in rock, pop, etc., have understood this and disregard it...
In "classical" or demanding music, etc., they pay attention to it, especially since it requires precautions given the potentially significant dynamic ranges...
The constant "rattle" about audibility... in the so-called "general population" is just excessively demagogic ;-)
OP
- Thread Starter
- #35
If I can clarify my reason for raising the question :
I keep seeing posts from the "general readership" questioning a particular devices ability to avoid ISOs - and clearly considering this is a problem worth rejecting said device over.
Example:
https://www.audiosciencereview.com/...ng-e50-review-balanced-dac.26219/post-2438208
Worse - it is now being used by marketing wonks, alongside jitter - as a justification for their overpriced hardware:
https://www.audiosciencereview.com/forum/index.php?threads/extreme-snake-oil.24765/post-2352433
I guess that is some sort of justification for including a test - so that we can demonstrate that lower cost devices also are immune (where they are)
Or by the misinformed, as an argument as to why DACs sound different:
https://www.audiosciencereview.com/...tra-streamer-preamp-review.57342/post-2164852
I keep seeing posts from the "general readership" questioning a particular devices ability to avoid ISOs - and clearly considering this is a problem worth rejecting said device over.
Example:
https://www.audiosciencereview.com/...ng-e50-review-balanced-dac.26219/post-2438208
Worse - it is now being used by marketing wonks, alongside jitter - as a justification for their overpriced hardware:
https://www.audiosciencereview.com/forum/index.php?threads/extreme-snake-oil.24765/post-2352433
I guess that is some sort of justification for including a test - so that we can demonstrate that lower cost devices also are immune (where they are)
Or by the misinformed, as an argument as to why DACs sound different:
https://www.audiosciencereview.com/...tra-streamer-preamp-review.57342/post-2164852
morillon
Major Contributor
- Joined
- Apr 19, 2022
- Messages
- 2,104
- Likes
- 510
While in most listening experiences it seems to be a minor issue, it's even embraced by sound engineers in mastering for the loudness war that drives sales*.. without considering "peak" levels, it would be interesting to observe the actual time spent in ISP mode...
on music with very low DR (Dynamic Range) and brought down to close to 0 dBFS... perhaps much more than can be considered... (The precaution taken, not new, for example, by RME with -2 dB is not accidental...such as the precautionary requests made by Spotify, etc.)
;-)
(* like conditioning, like normality, listening more and more "physio" )
on music with very low DR (Dynamic Range) and brought down to close to 0 dBFS... perhaps much more than can be considered... (The precaution taken, not new, for example, by RME with -2 dB is not accidental...such as the precautionary requests made by Spotify, etc.)
;-)
(* like conditioning, like normality, listening more and more "physio" )
Last edited:
As i most cases use EQ i end up making a -3db headroom anyway just in case . Can not an EQ filter in itself combined with some less than stellar source material generate an ISP
Would you provide more specifics on how you tested it? The Neutron HiFi DAC V1 on the current FW definitely does not include software volume control in its DSP,
Last edited:
OHtaru
Member
- Joined
- Oct 3, 2021
- Messages
- 71
- Likes
- 99
I connected Neutron HiFi DAC V1 to my Windows 11 PC, and played the 11025 +3dB file in WASAPI Exclude mode. The DAC output is connected to Babyface Pro FS IN3/4. Neutron HiFi DAC V1 is in 2Vrms mode, DSP off, THD compensation off. Ignore the grassy noise floor because of ground loop, I don't think this affects the conclusion.
Attachments
Where did you get the test signal? What player did you use? How did you lower the device volume?
Last edited:
