Thanks, got them.
Of course they sound different! The spectrum is different, the peak level is different, and so on. The only thing the same is the RMS level.
Attachments
Seems to me they are talking primarily about far-out phase noise, which is sometimes referred to as jitter by dac guys. Close-in phase noise may give different results. IME it will.
Really, only in audio forums???
Says who? What makes their opinions credible?
Beating again the same old dead horse. This is making up arguments without any factual support, it was discussed to death before. Anecdotical reports are not good enough.
I like you Markw4. Without you, life would be so boring
.The best performers are very young people, not those in their 70's or older. Given the average age of the members on this forum, it needs to be offset down for more realistic approach.
There wouldn't be so much controversy if posters (i.e. you) disclose the details of the listening test setup when posting claims of sound quality improvement by spending on boutique DAC or cables.
You may want to compare the wow and flutter magnitude of a turntable or tape with a clock close-in phase noise. Take the worse clock you can find around, its effect is still 4-5 orders of magnitude lower. If you still think you can hear this, you need to show evidence. Wait, you don't have any data, isn't it?
Yes, I made a similar point re: analog sources a page or so ago. I even reference the Manquen T2DS which is the best wow/flutter meter every made. It actually could have analyzed jitter too if it wasn't built for sources with much, much higher levels. High frequency flutter is an exact parallel with jitter, only far, far worse. And its baked into every analog recording to some degree, multiplied with every generation copy.
My only question is, why are we still talking about this?
Side effects of artificially induced marketing tactics. It's like our antibody reacting to pathogen intrusion.
Yes, the impact of jitter would be a function of the modulating signal (frequency, if a sine wave) plus modulation index. Without over-thinking this, the lower the modulating frequency, the higher the necessary modulation index to become audible. But that's also assuming a symmetrical, periodic sinusoid as the modulating signal, which should not be assumed as the classic jitter modulation type, primarily because of the root source. When we still did analog wow and flutter measurements, the result of the demodulator was filtered through a weighting filter, resulting in "Weighted WOW and Flutter" figures, meant to follow audibility better than unweighted. However the difference in mod frequency results in a major difference in perceived anomaly, with low mod frequency wow resulting in clearly audible pitch variation, where high mod frequency manifests as noise surrounding a "pure" sounding signal. The transition band seems to be between 100 and 500Hz, though it's not a hard/fixed transition, more of a slide or curve. But in general, low index, low frequency jitter and flutter/wow will be less audible. So different results, sure, but prehaps in a good way.
This has absolutely nothing to do with resitors, though.
I just read the wav files into Matlab and calculated the spectra. The signals are periodic so we can do an FFT without window. Then converted to dB and subtrated. The biggest difference is 0.005dB at around 1507Hz (one of the side bands), most likely due to 16 bit quantization errors. Note that windows before FFT is spreading the energy in the frequency domain (multiplication in time domain equals convolution in the freq domain).
- this may explain why you saw a larger difference. Without FFT window for such periodic signal, the spectral bleeding is absent.
The script is here:
[ya,Fs] = audioread('PureAM.wav');
[yf,Fs] = audioread('PureFM.wav');
N=size(ya,1);
%%
Ha=fft(ya
,1))/N;Ha(Ha==0)=1e-8; % avoid minus inf dB
[~,id]=max(dBa);
f=(0
N/2-1))'/N*Fs;f0=f(id)
dBa=20*log10(abs(Ha(1:N/2)));
Hf=fft(yf
,1))/N;Hf(Hf==0)=1e-8; % avoid minus inf dB
dBf=20*log10(abs(Hf(1:N/2)));
%compare magnitude spectra
idn=find((dBa>-100) & (dBf>-100) & (f>0));
[maxdifdB,idm]=max(abs(dBa(idn)-dBf(idn)))
f(idn(idm)),
plot(f,dBa);
axis([f0-500 f0+500 -80 0])
which echoes:
maxdifdB =
0.0050
idm =
7
ans =
1.5073e+03
would it be possible to address me in a more respectful manner?
I did not make claims about audibility levels, just pointing out a mechanism.
Another mechanism is that sidebands of a low frequency audio tone due to a low frequency jitter component are at low levels since the side bands scale with the tone frequency.
All I did was drop the files into Audition and run the Amplitude Statistics, and FFT on them. They are different peak levels, different envelops, entirely different waveforms. Why would anyone think the should sound the same?
There must be an error here. No other FFT has the spectrums identical.I just read the wav files into Matlab and calculated the spectra. The signals are periodic so we can do an FFT without window. Then converted to dB and subtrated. The biggest difference is 0.005dB at around 1507Hz (one of the side bands), most likely due to 16 bit quantization errors. Note that windows before FFT is spreading the energy in the frequency domain (multiplication in time domain equals convolution in the freq domain).
- this may explain why you saw a larger difference. Without FFT window for such periodic signal, the spectral bleeding is absent.
The script is here:
[ya,Fs] = audioread('PureAM.wav');
[yf,Fs] = audioread('PureFM.wav');
N=size(ya,1);
%%
Ha=fft(ya
Ha(Ha==0)=1e-8; % avoid minus inf dB
[~,id]=max(dBa);
f=(0
f0=f(id)
dBa=20*log10(abs(Ha(1:N/2)));
Hf=fft(yf
Hf(Hf==0)=1e-8; % avoid minus inf dB
dBf=20*log10(abs(Hf(1:N/2)));
%compare magnitude spectra
idn=find((dBa>-100) & (dBf>-100) & (f>0));
[maxdifdB,idm]=max(abs(dBa(idn)-dBf(idn)))
f(idn(idm)),
plot(f,dBa);
axis([f0-500 f0+500 -80 0])
which echoes:
maxdifdB =
0.0050
idm =
7
ans =
1.5073e+03 View attachment 1039529
Adjust your sensitivity to messages, there was nothing disrespectful in my message to you, it's all in your head.
If it's inaudible like you now seem to infer now, and you have no data or references to prove otherwise, then why are we even discussing about?
No need to. A FM signal of a carrier fo doesn't in general have the same period as an AM signal of a carrier fo, like your calculations are assuming. Only for small deviations the FM spectrum can be assimilated with an AM signal, otherwise, for large deviations, the spectrum trends to a continuum governed by a Bessel function.
Last edited:
That's the way of these threads. The OP usually disappears after a few days either shocked by the responses from Engineering minded types or because they were just trolling anyway. Sagen is doggedly hanging in for now though
Then things either drift completely off topic, or become a slanging match, or both. Then people get bored and drift off or the mods step in. Been repeating this pattern almost since the forum started.