Note that 4x oversampling can miss at most a 0.688dBFS ISP for a signal at 1/2 the original sampling rate (maximum valid frequency).
So, a -1dBTP is perfectly acceptable as a target:
That's what I'm saying. 4x oversampling helps but it doesn't solve the issue. Anything above 0, even if it's +0.001dBFS results in degradation of the original information. The rule is that digital audio should not be clipping, ever. It's not the same as analog audio.
This topic is bigger than annoying music production trends. It's about faithfully transducing digital information into analog audio. If there is an opportunity to improve such transduction, we shouldn't push against it IMO.
The original waveform either exceeds 0dBFS level in the analog domain, or it doesn't.
Nope. ISOs are invalid data because the input of the ADC has been overdriven.
One way to prevent ISOs is to add an analog peak detector at the input of the ADC which fires when the signal gets higher than the ADC is specified to accept.
This is correct. Leave enough headroom is the proper way to prevent them.
This is all completely wrong.
Virtually every professionally recorded signal is recorded leaving ample headroom to avoid digital clipping. dB meters are standard in audio interfaces, mixers and DAWs.
The reason why the digital samples get close to 0dBFS in finished audio products is what happens in the digital realm during audio processing: summing, gain boosting, compression, limiting, EQs, etc.
Even still, if the DSP/mix/master is done properly (which is most of the time, unless we're talking amateur production) none of the samples ever go above 0dBFS. Yes, even in the masters that are brickwalled to infinity.
With a "proper" (<0dBFS) master, the only time when samples can go past 0 is when an oversampling filter (or any digital filter, really) is applied, which is what happens in modern DACs.
Are you sure about that? My understanding is that ISO's are a random phenomenon due to the sampling theorem itself which only becomes a problem when sampling at or near 0 dB.
There is nothing random about this phenomenon. Every sampled wave, when reconstructed, goes past the sample points, except when the samples happen to coincide exactly with the peaks of the wave. When the samples get closer to 0dBFS, it's likely that the oversampled wave will run out of headroom and it will therefore get chopped.
Check earlier posts in the thread to learn more.
Are ADC's even used extensively besides vocals in most modern recordings? Seems like most of the "instruments" are just digital synthesisers? Could be wrong and trying to understand.
Yes, but it doesn't really make a difference. A properly digitally recorded signal and a digitally synthesized signal are identical in this context (they're all samples).
Once again, ISPs have almost nothing to do with ADCs or the recording/sampling process. It's a misconception.
The only time you can blame the recording/sampling step for ISPs over 0dBFS is when you record a signal that is almost clipping, then do nothing to it (no level adjustment, no compression, no nothing), then run it through a DAC or digital filter. The voice notes on your phone might have this issue.
6 dB can be spared easily as S/N of modern DACs is high enough.
Indeed. Lowering the volume before oversampling is the way to solve clipped ISPs. Ideally you'd have a dynamic system that lowers the volume just enough for a given audio track, but I don't know how one would go about implementing it. Surely it can be done in software.
On the practical topic of preventing intersample overs, do we know how the volume control is implemented on common devices such as Wiim streamers/amps or AVRs? Or is the sort of volume control required not commonly implemented?
The purpose of this thread is precisely to get reviewers to include ISP testing in their suites, because right now the majority's opinion is that ISPs aren't a big enough problem. That means we don't know how most DACs will perform in this area. I would say that they are an even bigger problem than sub-optimal SINAD, but that's just me.
docs.minidsp.com
