Nope, just hustling the suckers. His link demonstrates that he doesn't know what aliasing is, or he does and he's trying to confuse people who don't.
Charles,
As I have stated a number of times before, oscilloscope waveform pictures don't tell much about the spectrum of a signal. I suggest you get yourself, or borrow, a spectrum analyzer (they are not that expensive), or at least use a decent ADC and some spectrum analysis software.
Filter ringing is not distortion. As DF96 pointed out, "Causality requires a brick-wall filter to cause ringing, not by adding anything but by what is taken away - the 'ringing' was already there in the (artificial) waveform but was hidden by higher frequency components which the filter has removed."
hey what happened to charley ??
he demonstrates on his own website why you need proper antialiasing filters for adc's and reconstructions filters for dacs
check it out
Mother of Tone - The CD Format
is this dude just trying to take the **** out of everyone ??
For the love of music, I do everything
For those not familiar with the Gibbs phenomenon, it is a ringing that occurs with a signal consisting of a periodic signal (such as a sine wave) with jump discontinuities (such as by adding a square wave).
At this point it might be appropriate to ask Charles what it was that made him select a totally artificial test pattern that happens to be one of the best ones for demonstrating the Gibbs phenomenon...
please don't tell me that I did something 'right', if feels very uncomfortable
Would still love to hear what your rationale and thinking was for picking that specific type of test wave, if you don't mind?
Hi Julf,
for demonstration, I wanted a sine with a discontinuity, just to show that on the scope the sine is okay (there have to be filters in order to reconstruct a sine), and at the same time the step is also okay (without any ringing).
Probably only my Attraction DAC and Creation ADC are able to accomplish that.
But this implies a 2nd question: Why would I want to avoid the ringing if audio theory demands it for proper reconstruction ?
This journey began in the 90's when I started building some DACs for fun. The first 3 DACs were all sigma-delta, Crystal CS4334 (sucked), Crystal CS4390 (sucked), Burr-Brown PCM1716 sucked too, but had an adjustable digital filter, that sucked less, when programmed to slow rolloff.
Then I have read Burr-Brown's design seminar -> http://www.jitter.de/pdfextern/DesignSem5.pdf, which quite boldly stated, that sigma-delta DACs can never achieve the performance of their 'then' flaghip PCM1702 R2R/multibit DAC (This was the time before the industry put into service the AES17 filter, in order to improve the numbers on the data-sheets of all the following sigma-delta DACs).
So I made a DAC with the 'then' brand new PCM1704 24-bit/768kHz and DF1704 digital filter. This was the first DAC, that sounded decent and my first commercial DAC offering (Altmann Superlative 24/96 DAC, 1999). On the DF1704 you can also switch the digital filter between fast and slow rolloff, and slow rolloff always sounded much better to me.
As already shown here -> Mother of Tone - Time or Band, the slow filter produces less ripple on the step response but at the same time allows for higher alias distortion.
Now, its quite easy: If the filter with the shorter/less ringing sounds better compared to the filter with the longer/more ringing, then the next question would be, what would happen if the ringing would be eliminated completely ?
This was tried with the Attraction DAC, and I immediately fell in love with the sound. But not only I, it became quite a hit, and is still made the same way as it was introduced in 2004.
Somewhere in between I thought, okay, let's try another more modern sigma-delta DAC (I was fooled by the increasingly better data-sheet numbers, fooled by the AES17 filter). That was the DSD1792, and it sucked.
When I started working on the Tera-Player, I just tried to transport the performance of the Attraction DAC into a hand-held portable device. That was very difficult but after some trial and error it turned out very very good.
I just don't know what these old Philips DACs do to sound so good. In the meantime a Certificate of Excellence has been issued to Royal Philips B.V. for their TDA1543 and derivatives, and a 2nd Certificate to ARM for the new bit-manipulation instructions in the Cortex-M3.
Charles
There are no steps in music, because a step requires an infinite bandwidth. There may be sharp(ish) transitions, but these will excite less 'ringing' than a step. Note that the 'ringing' of Gibbs' phenomenon is quite different from the true ringing of a resonance, although they may look similar on a 'scope. Gibbs is about something missing, while a resonance typically boosts a narrow frequency range.
I suspect that Charles may be confusing sampling problems with ADC/DAC technologies. Shannon/Nyquist rules whatever technology you use. A particular technology may add its own problems, in addition to the unavoidable rules of Nature.
Now let me speculate why an aliasing ADC might sound OK on certain inputs, despite being wrong. The higher frequencies in audio are largely percussive sounds or modulated noise, plus some higher harmonics/overtones - although the latter will be at fairly low levels. At these upper frequencies we have little or no ability to discern pitch, so a burst of wideband noise around 20kHz won't sound much different from a burst of wideband noise around 24.1kHz. Therefore aliases do less harm than might be expected. (A similar argument may explain why filterless NOS DACs can sound OK, although there it is images not aliases).
What about signals at 40kHz, which will produce an alias at 4.1kHz where we can detect pitch? I think the answer is that most microphones or pickups are running out of bandwidth by then, as well as most instruments not producing much output up there. So a filterless ADC is OK provided your microphones and mixers are at the cheaper end of the spectrum?
Finally, note that whatever raw sampling rate is used in the studio if the music ends up on CD then the aliasing, if present, will be based on 44.1kHz sampling - plus any errors arising from interpolation i.e. you can't avoid Shannon by starting at 192kHz. Shannon/Nyquist is about the act of sampling, however it is physically implemented.
I suspect that Charles may be confusing sampling problems with ADC/DAC technologies. Shannon/Nyquist rules whatever technology you use. A particular technology may add its own problems, in addition to the unavoidable rules of Nature.
Now let me speculate why an aliasing ADC might sound OK on certain inputs, despite being wrong. The higher frequencies in audio are largely percussive sounds or modulated noise, plus some higher harmonics/overtones - although the latter will be at fairly low levels. At these upper frequencies we have little or no ability to discern pitch, so a burst of wideband noise around 20kHz won't sound much different from a burst of wideband noise around 24.1kHz. Therefore aliases do less harm than might be expected. (A similar argument may explain why filterless NOS DACs can sound OK, although there it is images not aliases).
What about signals at 40kHz, which will produce an alias at 4.1kHz where we can detect pitch? I think the answer is that most microphones or pickups are running out of bandwidth by then, as well as most instruments not producing much output up there. So a filterless ADC is OK provided your microphones and mixers are at the cheaper end of the spectrum?
Finally, note that whatever raw sampling rate is used in the studio if the music ends up on CD then the aliasing, if present, will be based on 44.1kHz sampling - plus any errors arising from interpolation i.e. you can't avoid Shannon by starting at 192kHz. Shannon/Nyquist is about the act of sampling, however it is physically implemented.
I shall be so bold as to mention that abraxalito has continued with the essence of what this thread was started for, on another forum. For those who are interested in the technical details of what he's exploring, and what he's up to, see here: Digital that sounds like analog
I'm pleased to see that he's making progress:
Improving "bloom" is shorthand for reducing that low level, detail masking disortion that digital reproduction is notorious for - he's making good progress ...
Frank
I'm pleased to see that he's making progress:
Improving "bloom" is shorthand for reducing that low level, detail masking disortion that digital reproduction is notorious for - he's making good progress ...
Frank
There's an old saying that applies here.
"A little knowledge is a dangerous thing".
Be careful which sheep you follow from the heard
what I found out from following that link?...apparently fas42 is banned over there so he cant participate..
we did several times mention how OT it was getting early on, but abrax seamed to think it worth the diversion... then Charles continued to 'defend' but has really just been digging that hole deeper and deeper.
I think weve explored that enough though dont you guys?
that mother of tine site is great comedy though...
we did several times mention how OT it was getting early on, but abrax seamed to think it worth the diversion... then Charles continued to 'defend' but has really just been digging that hole deeper and deeper.
I think weve explored that enough though dont you guys?
that mother of tine site is great comedy though...
Last edited:
So where does this thread go now? Do we turn it into a tutorial on digital audio, for the benefit of newbies and people hoping to sell equipment? Or would we end up going round in the usual circles with Shannon-deniers, Nyquist-deniers and Fourier-deniers all trying to claim that the usual rules don't apply to them, their ears and their wallets?
Well, we are at close to 300 posts in this thread - I don't think too many people have gotten this far. Sensible people dropped off early when they realized there was an impending wardrobe failure for royalty, and the cargo cult people dropped off (as proven by that other forum thread) when the objectivist trolls unfairly took over the thread from those willing to boldly go where no calculator has gone before...
If only the OP had presented his idea simply as a fun way to do an analogue implementation of a digital filter, with no claims about audible superiority, we might all still be playing nicely together. This is DIYaudio, so we can have fun provided we don't claim to be doing something deeply significant.
Well, at least one point that should be made is that there is no such thing as digital circuitry ... guess what? It's still really analogue, every last ounce of it, just pretending to be digital, there are no 0's and 1's anywhere in sight, all there really is just vast arrays of voltages bouncing around, and glitching merrily, but never quite badly enough to stop the circuits working as intending.
So, all we have to do is tell the sedate audio side of things to just ignore that hysterically cavorting digital crowd, under all circumstances ...,
Frank
So, all we have to do is tell the sedate audio side of things to just ignore that hysterically cavorting digital crowd, under all circumstances ...
,Frank
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.