A test of what can be heard in a room by people sitting in that room is best carried out by getting people to sit in that room and listening. The aspect being examined is quite simple: can people reliably hear a difference? If you find that they can hear a difference then you can explore what they are hearing. If they cannot hear the difference between unit A and unit B in that setup then the conclusion should be that in that setup those listeners cannot distinguish A from B. You cannot make A and B more dissimilar to check the threshold of difference because the test was of A and B, not degrees of A-ness and B-ness.mmerrill99 said:
The word 'test' does not necessarily mean 'carefully-controlled ethically-approved free from unintentional gender bias with sufficient statistics and full write-up suitable to pass peer review for an international research journal'. Did you never have a maths test or a spelling test at school? Maybe you objected to what the teacher called it?
It appears obvious that he is talking about the D to A step & that discrete is superior to the same step in silicon, no? Is this not what you also read him as saying with your 42 years of experience?
Perhaps this IMD is a sign of some deeper difference that isn't being exclusively measured in standard tests. Watts makes a point of mentioning noise modulation (which IMD would be related to) & the difficulties with measuring this but he does mention the audible effects of reducing noise modulation & those with Chord & Benchmark DACs may be able to compare & comment?
Again, the focus on a singular metric @ 130dB or the more incredible -300dB may be distracting
A "test" of scientific value (which many here proclaim Foobar ABX listening is & draw their conclusions based on this premise) usually entails a standard, reproducible setup & environment (remember the scientific method?) - what you are talking about is a listening session & the circumstances are significantly changed between sighted listening & Foobar ABX listening, yet you try to state that the results from one can be used to negate the results from another.
You are again falling into the misunderstanding about "ears only" listening - a phrase popularized here by SY - a great example of the same misconceptions
Last edited:
Literalism always progresses a discussion, don't you think?
Hoisting you with your own petard. An ESS dac is an ASIC, Chord DAC is an FPGA. both hunks of silicon. Both contain secret sauce algorithms to pump up the marketing bs . Functionally they both do the same thing. loads of maths and a low bit high speed converter. One sounds better as it appears to be the life work of unique genius as presented in print.
Ok, consider my petard well & truly hoisted, then - my eyes are watering
I was referring to the quoted statement by Watts which seemed to me to be about discrete Vs silicon conversion from D to A - to my mind the final step, not the handling of the digital signal prior to this step?
Last edited:
If you all agree that it has no scientific value then it has the same merit & value as sighted listening with which I agree & have stated already that "claims" should not be made based on either type of listening - they are simply anecdotal.
It could be argued that one has a high proportion of false positives & the other a high proportion of false negatives but we can't promote false negatives above false positives & believe only one is true because there is no scientific rigor about either
Last edited:
That is pure marketing gobshite IMO. The flipflops he uses are silicon and they come in little plastic chips. So how can they be anything other than silicon chips? If he made his flip flops out of individual transistors on the board he might be able to call it 'discrete' but he doesn't. he buys them in IC form. A small IC, but an IC nonetheless.
Are there any tubes made from CaF2? I can't see why they would be Otherwise good luck getting rid of silicon in any case.
I digress.
Have we had enough battles about effects, whether harmonic or noise or nonharmonic that are reliably detected at -100 dBFS? And by that I mean against music (if someone wants to listen to pure tones or ticks or noise all the time, be my guest but I'd hope no one would claim the latter translates to music well) I haven't seen any academic work make those claims. I'm being generous to give at least that wide of berth. Do we really need to talk about effects that are another 20-50 dB down ? Remember, in absolute difference that's 9E-5 to 9.9E-5.
Remember it has taken contrived test situations to get even a suggestion of audibility of effects 70 db down. Providing for 30 db of crest factor, which is again generous, this is the edge case for -100 dBFS effects. Can there be some acknowledgement that, so far, -100 dBFS is very hard to hear, if not beyond the threshold of inaudibility for a very large percentage of the audio enthusiast crowd. Anything lower than -100 dBFS loses even more prior plausibility.
I digress.
Have we had enough battles about effects, whether harmonic or noise or nonharmonic that are reliably detected at -100 dBFS? And by that I mean against music (if someone wants to listen to pure tones or ticks or noise all the time, be my guest but I'd hope no one would claim the latter translates to music well) I haven't seen any academic work make those claims. I'm being generous to give at least that wide of berth. Do we really need to talk about effects that are another 20-50 dB down ? Remember, in absolute difference that's 9E-5 to 9.9E-5.
Remember it has taken contrived test situations to get even a suggestion of audibility of effects 70 db down. Providing for 30 db of crest factor, which is again generous, this is the edge case for -100 dBFS effects. Can there be some acknowledgement that, so far, -100 dBFS is very hard to hear, if not beyond the threshold of inaudibility for a very large percentage of the audio enthusiast crowd. Anything lower than -100 dBFS loses even more prior plausibility.
I don't know the details of the Chord DAC he is talking about but you seem to know, so I have one question - The FFs are handling digital signals not doing D to A conversion - what is after the FFs that does the I to V conversion, silicon or discrete resistors?
YouTube
Then read this Chord Electronics DAVE D/A processor | Stereophile.com the biggest handwavy stuff is over the implementation of the 'analog noise shaper'. which is the gain stage after the conversion. Nested feedback loops with an ADC sampling to feed back into the digital processor. hundreds of dB of feeback. The NOS DAC crowd will be cowering in a corner!
IV conversion appears to be resistive. But in my book the current is still analog so the flip flops ARE the converter.
Then read this Chord Electronics DAVE D/A processor | Stereophile.com the biggest handwavy stuff is over the implementation of the 'analog noise shaper'. which is the gain stage after the conversion. Nested feedback loops with an ADC sampling to feed back into the digital processor. hundreds of dB of feeback. The NOS DAC crowd will be cowering in a corner!
IV conversion appears to be resistive. But in my book the current is still analog so the flip flops ARE the converter.
IMHO ...... many people when they detect a difference between A and B, go to the THD/IM test to see what differences exist there. And, then attribute what they detect to those measured numbers. When in fact, those tests often have nothing to do with what they heard/detected. And, those tests may show very low static distortion numbers/differences.
THx-RNMarsh
THx-RNMarsh
You need to be a bit more specific, do you mean -60 dB below the average program level or the instantaneous level?
When measuring with a sound level meter or even a VU meter you are getting a time weighted average and it is certainly possible to hear artifacts 60 dB below that level.
If you are looking at instantaneous signal voltage levels, then it must be something that becomes larger with perception based weighting to reach a perceptual level.
Ordinary harmonic distortion would only be masked for those with human hearing. I wonder about some of the others making claims.
As an example I recently had someone with an issue that the mixer they were using had built in digital effects processors. In a quiet room the background noise produced by the effects processors could be heard. When the musicians were performing the constant noise was masked. So we really should be talking about instantaneous signal to noise ratios not absolute maximum output signal to noise ratios. We also should be sure to distinguish between static conditions and in use conditions.
Last edited:
FFs definitely work in the digital domain & are not the analogue converter, no matter what individual "book" or frame of reference you have.
The bolded part is what I assumed Watts was talking about by discrete?
- Status
- Not open for further replies.