You asked for "proof that it's possible for a dac to differentiate a sawtooth wave from a sine wave from a couple samples", which is exactly what I posted. If this is not why you joined the thread, then why did you make the false assertion that it's not possible AND then asked for proof that it was?
OK, you're disengaging because you got the proof you asked for, which falsifies your assertion.
Thank you for your service and intent to make me better understand these things.
I have my particular reasons. For now I'm just interested in practicalities of how to make my gear perform as I want it.
I like the discussion, although most goes over my head. Maybe the issue between players is an effect that is getting applied by one software or the other.
An ASRC, by its nature, has a clock going in, in this case 44.1kHz word/frame clock and the associated Bit clock, and the outgoing clock, at 48kHz and the different Bit clock
If an ASYNCHROUS sample rate converter has the same clocks, it is not converting anything. A SYNCHROUS sample rate converter is converting multiples of teh same clock, like 48kHz to 96kHz, then is can work from the same master clock, and derive the bit clocks, but this is not what we are talking about. ASRC: Its entire purpose is to change to the different clock speed.
See above
Show me where they are peanuts for a stand alone ASRC:
That is without jitter. The datasheet says -102dB THD at 1kHz, but this is just the arithmetic errors. The datasheet also says a loss of 1dB of degradation at 10nS jitter. Still fine, but the new ones do better.
It is. However they don't care very much. How do I know? A world class DSP engineer I have know for 30 years worked at CSR when tehy were bought by Qualcomm. I trust him over someone who has to be right all the time on the internet.
J
Why can't we convert the sample rate with the same clock? For example, let's say we have an input sample rate of 48kHz and an output sample rate of 44.1kHz. We get a clock that oscillates at 7,056kHz (or from which we can derive 7,056kHz) then divide that by 147 to get the 48kHz sample rate and by 160 to get the 44.1kHz sample rate. Same clock, two different sample rates. This is just an illustrative example of course, even old clocks operate at far higher frequencies.
But the purpose is not to change to a different clock speed, we want the clock to be the same speed, the purpose is to change the number of samples per (the same) second.
Firstly, we're not talking about a standalone chip but a SOC and more importantly, we're not talking about hardware SRC but software SRC (Android). So most of the above is irrelevant anyway, because there is no clock, there's just data and processing, time is irrelevant.
That's just an "appeal to authority", an unacceptable fallacy in this sub-forum. Likewise, your relative level of trust between me and your friend is both unimportant to me personally and again, irrelevant in this subforum.
You are also forgetting: There is latency, so the jitter affects different samples.
There is no reason to assume they have done the job properly. There are multiple citations that they haven't, and like you I want proof. Having worked with Qualcomm, and measured in detail their attempts at audio (have you seen what they do to AAC? 14kHz brickwall bandwidth in many cases. Definately not transparent.) I side with them not being able to do a good job, or being bothered.
There's two serious issues with this assertion:
I'm not forgetting latency but maybe you're forgetting that latency doesn't affect jitter? Latency is the period of time required for (SRC) processing, which can be up to 30ms or so. However, that (say) 30ms processing delay is applied to all the samples, so there is no added jitter.
Yes, there is. SRC is a common, widely used process that's accomplished transparently as standard by everyone else with modern processors and even free off-the-shelf algorithms/libraries. The freely available MatLab resampler code for example has artefacts down at around the -170dB level! So, it's not so much about having "done the job properly", they'd pretty much have to deliberately screw it up to get audible artefacts. But again, my understanding does not go deep enough to be absolutely certain.
1. There are also multiple citations that silver cables sound brighter than copper cables, doesn't make it true though. There are numerous similar examples, which is why anecdotal evidence is considered amongst the very least reliable types of evidence. Even if the anecdotal evidence is correct, we still have to ask "when?". The probability of Android not performing audibly transparent SRC is far higher in early devices and early versions of Android.
But the AAC format is supposed to have a low pass filter set at about 14kHz (below a certain bit rate). Additionally, as I'm sure you know, Android is developed by a consortium of dozens of companies. So how do you know that it was Qualcomm who developed the SRC code in Android? Even if it was Qualcomm AND they screwed it up big time, how is it that none of the other companies in the consortium noticed or did anything about it?
Again, ALL of that is false (or can be false), because you don't get to define "normal use" for the whole of science. Even your assertion about "life or death" is false, it CAN be a matter of life or death because most of the digital signal processes used to record and reproduce "music for pleasure" are not designed specifically for "music for pleasure", they're designed for digital audio in general or for other A-D-A applications. For example, a wearable heart monitor can be a matter of life or death and medical scans/images certainly can.
Of course they do, that's largely what science is! Didn't you know that? For example, the science of physics (inc. Sound and audio) is based on theories and laws; "rules" that are true within a set of conditions. Finding an exception within those conditions invalidates that "rule", it's what stops a hypothesis becoming a theory or proves that an existing theory is wrong (or at least incomplete) and of course vice versa; the lack of an exception is what allows a hypothesis to become a theory.
Yes you are in a Science forum if you reference "people" in general and even more so if you specifically reference science! (quote: "It’s as prevalent in science circles as it is among subjectivists.")
No you're not, because ....
As already explained; "laboratory quality" is required because without it the multiple SRC processes on some/many of the recordings you probably play in your living room would have audible artefacts and "I would venture to say that the vast majority of the rest of this forum" would also be able to hear them!
No, I'm not an exception in this case, because: Firstly, I have had various consumer systems at home over the years and Secondly, you wouldn't need a professional system to hear the artefacts of cumulative multiple SRC processes performed by an SRC that was only just inaudible, it could/would be audible in normal home use.
I do consider the audience of hobbyists (and the general public), that's what I do for a living!
And in this thread too. Clearly you don't know the difference between a hardware DAC and a software SRC process or the difference in their application?
