The Scretary of War and Finance drug me to another Estate Sale last weekend and while stumbling through rooms full of trash with price stickers on it, I found a Studer Revox Reel To Reel in mint condition for $100 and immediately bought it. Took it home cleaned it up, works perfectly. Recorded some vinyl to it and sounded wonderful but with the usual clicks and pops. Then while listening to a couple of Hi-Res downloads on my computer got the hair-brained scheme to use JRiver to play the files bit perfectly to the DS and then analogue to the Revox. Rewound the tape, hit play and the sound was stunning…warm and full like analogue, no background clicks or pops, convenient source selection like digital. It sounds better to me than CDs thru the DS, better than the Hi-Res thru the DS (I’ll bet the forum wizards can tell me why) and I can build a 2 or 4 hour mix tape and never have to set my drink down to change discs or Records or stare at a computer screen. And I can insert my PSA GCHA anywhere in the chain for a different sound. I love the DS even more now and have free wow, flutter and hiss. I played it for another audio guy and he pointed a finger at me and said “see, that is 100 times better than computer music or digital junk”. I didn’t have the heart to tell him…
Great find and wonderful fun.
You shoulda told him.
We do like that analog magic. Tape adds several artifacts to the sound that many of us find pleasing. I am listening to DSD transferred from the original high res PCM by Gus Skinas. My jury is still out on whether there are actual differences but I wonder if any heard between the original PCM and DSD are related to even order harmonics added by single rate DSD? Elk?
Yes tape has its own sound, particularly the gentle compression it adds as it reaches saturation.
I am not aware of DSD adding harmonics.
Single rate DSD is better at representing a 10kHz square wave than Redbook PCM. In doing so, it avoids adding first and third harmonics as does PCM, but this is much more theoretical than audible. The filters used, etc. has a much greater impact on audibility.
PCM converted to DSD should not sound better than the original PCM. However, a good transfer to DSD certainly could sound better than a less than pristine original release of a PCM recording. I suspect this is why a lot of “remastered” releases sound better than the original; the new version was simply treated better.
Yes, it can be a bit difficult to sort out the actual path that the tracks took to your door. At least Acoustic Sounds is pretty good at giving you a peek at that.
For me, there are so many variables that have greater impact on the sound than format used. Both sound great if good equipment, properly set up, was used.
I am interested in your comment of DSD adding even order harmonics. I have not understood this to be the case.
Wglenn, I read that about even order harmonics and DSD as well. I will try to find it. The contention was that the harmonics interacted with the shape of the human ear but I can’t remember if that process converted them to even or if the DSD process did. The statement was made that digital can NEVER sound right as our ears and neural circuits have evolved to decode analog and digitization produces an unnatural waveform.
Well, I can’t find serious evidence of this and it may have been an offhand comment that stuck with me. I thought that it was perhaps from aliasing of high frequency noise… no. Maybe from ringing due to steep low pass filtering (stop laughing, Ted)… no. It is clearly not the result of quantization error or jitter. DSD, done right, suffers from vanishingly low levels of non-linearity. It is there, but on the order of -150 dB and beyond. So far, looks like a dead end.Ted?
@Stevie B: We love our even ordered harmonic distortion. It is very obvious in the case of heavily overdriven tube circuits (Jimmie Hendrix) but less so in the use of small amounts of harmonic distortion in the studio. As I learn more about the techniques used to capture audio I realize how much “interpretation” goes on by audio recording engineers. You are (often) not listening to anything that resembles the original product. Processing is done to get that initially recorded signal to sound more pleasing to the ear, even if that means getting it to sound more “natural”. One of those tricks is adding tiny amounts of harmonic distortion using minimally overdriven vacuum tubes or, lately, software emulation of this sound. Harmonic distortion… you are adding tones above and below the fundamental frequency that enriches the sound. It is a “little white lie” that we usually find pleasing when used in reserved amounts. It is not necessary by any means but many raw recordings are flawed by the techniques and equipment used (or misused) so that fudging the sound makes them more attractive. There are many other tricks used. The only way to know what it really sounded like was to have been there when the recording was made.
Stevie B said The statement was made that digital can NEVER sound right as our ears and neural circuits have evolved to decode analog and digitization produces an unnatural waveform.I have seen this argued both ways: the ear is analog, because it is a natural part of our world and must therefore be analog. So there.
Or the ear is digital: sound stimulates the hair cells of the epithelium which leads to the depolarization of the receptor cells of the Organ of Corti which creates an electrical nerve impulse which travels the spiral ganglion sending information through the vestibulocochlear nerve to the temporal lobe of the brain which interrupts these pulses as sound. The ultimate digital system. Therefore digital reproduction sounds best.
Both arguments strike me as nonsense. Each adherent starts with the premise his preferred method of reproduction sounds better and works backward to claim this preference is physiological. Neither proponent explains why so many prefer the opposite method of recording and playback.
Excellent, educational responses Gentlemen. Jeez, I feel a little betrayed by the recording, mixing, mastering process.
Here is an interesting take on the debate:
If the premise of the article is true, then Ted may be able to add the required “artifacts” during D to A conversion. Another slot could be added to the DS and a Analogizator board inserted. An additional Menu Item for “Vinyl/Tape/Tube Amp/Off” could be added and Voila´ perfect analog reconstruction. I wonder if the artifacts would survive the BHK amplification chain…BK could fix that. I’ll bet Paul would love this idea.
Also, did you notice the reference to “Western” ears that enjoy specific distortion artifacts. If there are a pair of “Eastern” ears reading this, I would love to know your position on comparison of the two processes. That preference may be cognitive.
A nice little article. I am pleased he makes the point that digitally sampled waveforms contain all the information of the waveform. Many believe the waveform is “stepped” when digitally recorded and played back. Untrue. The playback is a normally shaped analog waveform. (Of course, every recording has some level of imperfection/distortion.)
With respect to various harmonic distortions/additions sounding better to at least Western ears, I believe the writer is being careful to state this may not be universal. Eastern music does employ scales different from Western, and often recognizes quarter tones, etc. Equal temperament, used in tuning a piano which makes all 12 steps of an octave scale equal intervals, makes music in all keys similarly out of tune. The advantage is that the piano need not be retuned to change key. Eastern music uses different intervals and does not employ equal temperament. Our music sounds out of tune for a classical Eastern musician. Of course, we have difficulty grocking Eastern music as well. (As an aside, classical string and wind players, and singers, perform in pure temperament unless accompanied by a differently tuned keyboard.)
Recordings are not automatically sweetened with harmonic additions. This is simply one tool in an engineer’s toolbox. As one would expect, pop/rock is heavily processed and manipulated. The end result often does not sound as it was recorded. (Similar to the dreadful photo processing many use when posting to the Internet - do people really think these photos look good?) Classical recording is rarely compressed, never sweetened/processed, and usually is usually just subtly EQ’d to enhance a sense of naturalness.
Stevie B said Another slot could be added to the DS and a Analogizator board inserted.I forgot to mention that such processors have been sold for years in the form of tube buffers/tube output stage; e.g., click. These are designed to add pleasing tube distortion which fattens and sweetens the sound. This can be replicated digitally as well with plug-ins but does not have the same emotional appeal. :)
Ironically properly implemented digital is smoother than analog - what it’s missing in digital are the higher frequencies (the ones above the antialiasing filter cutoff in the A/D converter.)
As to simple distortions like 2nd or 3rd order distortions: in general a truly balanced implementation of analog will, in theory, nuke all deterministic even order distortions. Conversely many digital processes can easily introduce odd order distortions.
In my view the problem with digital is that there are a lot more places for unnatural distortions to show up. I.e. high frequency hash from the sharp edges of digital signals, clock jitter, groundloop hell from the many boxes we use in digital setups, etc. With pure analog processing the distortions are often more benign since they are likely to resemble the distortions that we are subject to in the real world already, and therefor already discount: things like reverberation, echos, changing FR depending on distance, position, etc.
Back to DSD distortions: Most of the papers that talk about SDM converter’s and their distortions are computer simulations of essentially chaotic processes. We don’t have good enough theoretical models of the sigma delta process to calculate or predict most distortions any more efficiently than doing a simulation and measuring the output. But these results are very specific to the particular SDM being modeled and those models almost certainly are different than the SDM implementations actually being used in the real equipment. In the case of the DS where the SDM is done purely digitally I consider the specifics of how I (mostly) avoid noise modulation and overflow to be a trade secret and that there’s no plot out there that will show any particularly simple distortion, e.g. second order distortion… of the processes in the DS. Since I use balanced technology in the DS there’s little 2nd order distortion above and beyond the output transformers. I can see 2nd and 3rd order distortions on the scope but their levels are at essentially inaudible compared to the levels of euphonic distortion people are usually talking about for tubes, speakers, etc.
Thank you, Ted! Great stuff
Analog is additionally far from flat in frequency response. The bottom end is particularly problematic, with peaks, dips, all sorts of oddities.
While digital filters do impinge on high frequency response, analog often has big problems at this end as well with more peaks and dips. I am hard pressed to opine whether digital or analog is better in this respect.
Yes, Ted, I can see where you have lots to play with in the FPGA code, although I understand very little of the actual process. Probably less than I realize.
I thought that this was a decent primer.
@Stevie B: I have very little experience with “things recording”, but hope to gain more as I go. When I have tried to ad a bit of “tube sweetness” to a track I have ended up not liking it compared to the original. Perhaps I use too much or apply it incorrectly. It certainly is not a panacea but I understand that it is used quite a bit in pop recordings. Other forms of distortion are used, too. Clipping (ick) and extravagant compression are two other types that I dislike immensely.
wglenn, Your comments concerning the recording process are very interesting. I guess I am naive but always thought the process was kept as pure to the original as possible. Maybe we blame the hardware and software for unless than perfect sound when it has actually been manipulated during the trip from the studio to my equipment rack. Here is a perspective from a recording engineer on analog vs digital and if you search his blog, you can find some interesting discussions on loudness vs dynamics.
I have a much better appreciation for what Paul, Ted and Arnie must contend with in trying to accurately reproduce instruments and the human voice with ever-changing technology.
There are those who level the field as tape , then vinyl and a distant third is great digital . I for one see things as much more complex than that. First off most here have one system maybe two , as I have two speaker setups and many headpones our beloved music has many faces . I think what you have found is that change is good meaning you just like the sound of your music processed as you did. Clearly it’s not as close as the original but yet you still like it. As we look for purity in our audio it does not lead me to what I like always. How many even use any form of eq , or processing on our end ? I would guess most do not , my point is as I do not either , having a few setups shows me there is other views in the window of sound we listen to. Many years ago I had a set of Allison acoustic speakers they had a set of switch son the back the changed the sound of them . As I was not happy with there sound I called up the company. I was lucky to talk although the owner . One of his comments was why do you not use the bass, mid and treble on my preamp ? That lead me to buying an eq . As that road has many turns it still leads you to some pretty good places at times. I a, willing to bet your results will not be pleasing worth all recordings , does it ?