Exact fractions, with no end effects? A sustained note would retain the same waveform all along? For this to happen you would need a pipe with exactly the same cross section all the way along (e.g. no narrowing when you get near the blowing end), no open end, no acoustic impedance conversion to air etc. A string would need to go from completely free to completely clamped at the ends, at a single point.cbdb said:
I have decided to make "better" measurements (= using more clean sound card) of distortion profile of the amplifier under test. Attached are measurements at 0.5W, 3W, 6W and 18W / 4ohm power. As a source there was a 0dBFS sine wave in 44.1kHz/16bit format, supplied by a CD player and DAC, which is pretty clean itself. There is a stepper attenuator at the amp input. We can see, that the dynamic range of the hybrid amplifer under test is better than that of the 16-bit file, as can be seen from the noise floor decay above 22kHz. One can also see decaying distortion of the amplifier as a function of decreasing level.
Attachments
Last edited:
Hi Pavel.
I have taken a listen to apricot.wav and avocado.wav files and I do hear differences from my usual listening position at my PC desk 90* off axis to my speakers.
I note that avocado sounds subjectively louder than apricot, I note that avocado has a 'bursting' distortion quality that causes a subjective stacato/gating effect causing false dynamics and I note that avocado whilst having more mid/highs has less information than apricot ie avocado suffers some masking.
I also find that listening to either track directly from the Foobar playlist is distinctly different to playing either track in the ABX applet.
There is a strong sameness masking noise over both the ABX tracks that alters and diminishes track differences that are easily apparent when playing direct from the playlist.
In other words Foobar ABX applet is essentially USELESS and will force false negatives when testing even relatively coarse differences.
I have run multiple rigorous testings of this assertion and have well proven this finding to my satisfaction.
Sorry to rain on the parade, but this is what scientific method is about.....be certain that the testing methodology is correct elsewise testing results will not be correct.
In answer to your poll questions.....
I presume avocado with level dependant distortions and odd dynamics is the tube amp, apricot the original.
I do not like either recording in particular because of the vocal extreme limiting/overload in parts, despite this I find apricot less objectionable.
Avocado on first listen is more 'impressive' but for me in short order this file with lack of detail information and weird dynamics becomes annoying and tiring to listen to, YMMV.
This Foobar configuration in conjunction with hotkeys Before and Next and mouse click anywhere in the progress bar allows me to flip between tracks seamlessly, instant (repetative) AB sighted testing.
Dan.
I have taken a listen to apricot.wav and avocado.wav files and I do hear differences from my usual listening position at my PC desk 90* off axis to my speakers.
I note that avocado sounds subjectively louder than apricot, I note that avocado has a 'bursting' distortion quality that causes a subjective stacato/gating effect causing false dynamics and I note that avocado whilst having more mid/highs has less information than apricot ie avocado suffers some masking.
I also find that listening to either track directly from the Foobar playlist is distinctly different to playing either track in the ABX applet.
There is a strong sameness masking noise over both the ABX tracks that alters and diminishes track differences that are easily apparent when playing direct from the playlist.
In other words Foobar ABX applet is essentially USELESS and will force false negatives when testing even relatively coarse differences.
I have run multiple rigorous testings of this assertion and have well proven this finding to my satisfaction.
Sorry to rain on the parade, but this is what scientific method is about.....be certain that the testing methodology is correct elsewise testing results will not be correct.
In answer to your poll questions.....
I presume avocado with level dependant distortions and odd dynamics is the tube amp, apricot the original.
I do not like either recording in particular because of the vocal extreme limiting/overload in parts, despite this I find apricot less objectionable.
Avocado on first listen is more 'impressive' but for me in short order this file with lack of detail information and weird dynamics becomes annoying and tiring to listen to, YMMV.
This Foobar configuration in conjunction with hotkeys Before and Next and mouse click anywhere in the progress bar allows me to flip between tracks seamlessly, instant (repetative) AB sighted testing.
Dan.
AB sighted testing? I assume the sameness masking noise is unmeasurable. There must be an SDK for foobar somewhere, I'd put my money on it calling the same code as playing from the play list.
Last edited:
Hi Dan,
thank you for listening to the files and for your comments on their sound. I will disclose which is the original and which is recorded through tube hybrid amp after December 4, when the poll expires.
A disappointing turnout so far, I will therefore give them a good listening to and vote. I have no means to use Foobar having only a chromebook, so you will be pleased to know it'll be completely unscientific, but 100% subjective and biased towards not hearing a difference, because I don't expect to for a number of reasons that I won't bore you with, and that you can probably guess at anyway.
Take a closer look at the screenshot and you will see my method.
By pressing Before or Next hotkeys (need to be custom configured in settings) and in this case keeping the mouse pointer at 1:18 (or any particular point in any track) I can instantly compare playlist items repeatedly in any sequence...ABABAABBAAABBBABBA etc.
With practice and suitable program material discerning fine differences by this method is reliable I find, and correlates with longer term listening findings.
I can then pause the playlist (spacebar) and mouse click on the ABX applet Play A or Play B buttons perfectly easily and quickly.
I can then Stop ABX operation and return to the playlist perfectly easily and quickly.
This method allows near instantaneous playback comparison of all four playback conditions.
Scott: There is one difference and that is the ABX applet writes to an AppData\Local\Temp directory whilst initialising....perhaps reading from this Temp directory causes something untoward.
The difference I'm hearing is a broadband dirty noise/distortion cast over the sound obscuring detail and setting an overriding sameness to the test files.
I suppose it ought to be measurable.
Bill: I have stated this Foobar problem in the past, I have better sound system now and the problem I state is readily audible I find still.
If you don't go from the ABX applet back to the playlist you will never hear or know about the problem.
I got curious in the past when I found I had trouble picking differences in ABX that I could reliably hear in the playlist whilst practicing immediately before ABX testing.
Go ahead, try my method and take a really critical listen to Foobar itself before listening for differences in files.
Dan.
Last edited:
No, I have encountered this on other PC systems.
And in any case here the only variable that I know of is the ABX Temp file operation within Foobar.
It would be good if you can measure a difference, that would be very interesting/useful.....if Foobar is proven to be defective as I suggest then so be it, we need a different ABX solution or workaround.
I suppose I could try running ASIO on this laptop, but if memory serves I was running ASIO on a different laptop when I first noticed this issue.
Dan.
And in any case here the only variable that I know of is the ABX Temp file operation within Foobar.
It would be good if you can measure a difference, that would be very interesting/useful.....if Foobar is proven to be defective as I suggest then so be it, we need a different ABX solution or workaround.
I suppose I could try running ASIO on this laptop, but if memory serves I was running ASIO on a different laptop when I first noticed this issue.
Dan.
Last edited:
Lacinato ABX. Vade retro Direct Sound.
Dan,
Lacinato ABX
ABX/Shootout audio testing tool
Question: foobar2000 with Direct Sound, Wasapi, ASIO, Kernel Streaming... which?
foobar2000: Components Repository - Kernel Streaming support
foobar2000 ABX plugin and Lacinato ABX work with Direct Sound.
******* *******
Note: Direct Sound to play music, in my very optimized W10 Pro 1709 64bits, is like vade retro satana. Well, at least, since W XP.
Vade retro satana - Wikipedia
Dan,
Lacinato ABX
ABX/Shootout audio testing tool
Question: foobar2000 with Direct Sound, Wasapi, ASIO, Kernel Streaming... which?
foobar2000: Components Repository - Kernel Streaming support
foobar2000 ABX plugin and Lacinato ABX work with Direct Sound.
******* *******
Note: Direct Sound to play music, in my very optimized W10 Pro 1709 64bits, is like vade retro satana. Well, at least, since W XP.
Vade retro satana - Wikipedia
Dan, I will check it. The 24-bit loop should tell if there is an issue.
I am sure that you know it, but for the others, when playing sound under Win7 and higher and you want to utilize 24-bit resolution, please do not forget to set 24-bit resolution for the soundcard used in the Control Panel, otherwise 16-bit is default and hirez files are converted to 16-bit resolution.
After I read Maty's comment on Direct sound, it is necessary to make this setting!
I am sure that you know it, but for the others, when playing sound under Win7 and higher and you want to utilize 24-bit resolution, please do not forget to set 24-bit resolution for the soundcard used in the Control Panel, otherwise 16-bit is default and hirez files are converted to 16-bit resolution.
After I read Maty's comment on Direct sound, it is necessary to make this setting!
Attachments
Dan,
I am sorry but I cannot support your hypothesis, at least not on my system. I have recorded 24-bit test files first played by Foobar2000 directly, second by Foobar ABX plug-in as A, B, X and Y. Always recorded by SpectraLab SW. In all cases, the recorded files are true 24-bit and they are same in max rms, min rms, avg rms, differing only in an analog instantaneous noise value, which is of course the random variable. They do not differ in rms noise level, but in instantaneous values, which corresponds to DAC - ADC inherent noise and in a 24-bit mode this is far below CD 16-bit noise bottom. So nothing, the difference is same as two DAC-ADC runs without signal or two DAC runs when you make your sighted test.
P.S.: I made a Diffmaker difference as well and all I got was a pure noise, same level as residual channel noise.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.