JohnnyCanuck
Headphoneus Supremus
Oh, good. You found the abridged version.
JC
JC
What a long, strange trip it's been -- (Robert Hunter)
- Thread starter Baldr
- Start date
audio philestine
100+ Head-Fier
I listened to this carefully and thoroughly enjoyed the discussion of ribbon mics. When they got to the discussion of DSD they seemed to get a little glassy eyed and began invoking a lot of anecdotal evidence, thought experiments and speculation.
Some standouts were the discussions around 10k limits, orchestral string sections, 20k square waves and conversion of DSD to other formats.
There was an interweaving of arguments that in natural settings there is very little above 10k due to normal hearing, and losses in air, yet ultrasonic elements are important and that string instruments sound awful close up due to high frequency components.
Strings do not sound awful close up. One of my favorite albums is Vilde Frang Homage where the recording is so clear and up close you can year the bow scraping on the strings before they begin to vibrate. Rather than wanting to rip the headphones off, it's a totally beautiful and immersive experience.
They invoke a thought experiment about 20k square waves sounding different that 20k sine waves as demonstration that ultrasonic tones are relevant. They don't seem to have done the experiment and there's no discussion of the physical limits of transducers.
There is a rather remarkable statement that PCM can't accurately capture higher audio frequencies at any sample rate.
There is a reasonable argument for higher sample rates and more gradual high frequency roll-off than you get with 44.1k sample rates.
There is some very non-scientific speculation about interpolation and zero-crossings that seems blended with concern for dithering.
They start to invoke magic when claiming that something recorded with DSD then converted to PCM (and even further converted to MP3) will sound better than a direct conversion to PCM. If there is any truth to this then it would be more of a comment on the quality of modern ADC's rather than innate limitations on PCM or innate superiority of DSD. Again, though, their comments are general and anecdotal. The do cite a recent recording session they did where they recorded with a single mic to a DSD recorder and multiple mics to PCM, later mixed down. They note that the two methods aren't directly comparable and they also don't note which result they liked better (though one could assume) or offer characterizations of the two results. They do say they are planning another recording session taking the same analog signal and recording via DSD and PCM for a more direct comparison, but that hadn't been done yet (The video was made October 2018).
They attribute to DSD many of the characteristics and qualities people seem to attribute to listening to music via a Yggdrasil
I also watched this, and while I also felt that some of the technical arguments came close but fell a bit short, there were a few take-aways
1) They are musicians (and a recording engineer) with well trained ears that tried a lot of equipment and compared and found a sound the liked, even if they can't explain the science in a way that I can buy into. They're hearing something, even if I can't explain it.
2) I interpreted their comments on the DSD recording sounding better when mixed down to mp3 as more a commentary on a short, simple, and refined recording chain. Very nice mics placed with thought around good musicians in a nice acoustic space, into nice mic pre-amps and direct hi-res digital recording. No mucking around with mixing or effects or questionable post processing techniques. Quality recording makes for a better end product, even after downconverting,
While DSD doesn't get any love around here, [personal opion/soapbox to follow; I'm sure not everyone will agree...] I get the impression that what they liked about the DSD recorder it is that it takes the internal oversampled representation (e.g. 1bit at 2.5+ whatever MHz) of the music internally in the ADC and records it directly, without attempting to downsample it back to PCM (24 bit, 96,192, whatever), and then feed it to an oversampling DAC that converts it back to the 1-bit MHz+ internally before making it back into music. That's a lot of extra conversion steps which historically have not been done well, so in theory eliminating those conversions should make for a better end product. But... most of us here have bought into the magic of baldr's expertise at doing PCM very well, rather than DSD's attempt to perform a technological run around. DSD assumes doing less is better than doing something poorly, but we've discovered doing something well is better still
ehayduk
New Head-Fier
- Joined
- Feb 3, 2011
- Posts
- 17
- Likes
- 27
I listened to this carefully and thoroughly enjoyed the discussion of ribbon mics. When they got to the discussion of DSD they seemed to get a little glassy eyed and began invoking a lot of anecdotal evidence, thought experiments and speculation.
Some standouts were the discussions around 10k limits, orchestral string sections, 20k square waves and conversion of DSD to other formats.
There was an interweaving of arguments that in natural settings there is very little above 10k due to normal hearing, and losses in air, yet ultrasonic elements are important and that string instruments sound awful close up due to high frequency components.
Strings do not sound awful close up. One of my favorite albums is Vilde Frang Homage where the recording is so clear and up close you can year the bow scraping on the strings before they begin to vibrate. Rather than wanting to rip the headphones off, it's a totally beautiful and immersive experience.
They invoke a thought experiment about 20k square waves sounding different that 20k sine waves as demonstration that ultrasonic tones are relevant. They don't seem to have done the experiment and there's no discussion of the physical limits of transducers.
There is a rather remarkable statement that PCM can't accurately capture higher audio frequencies at any sample rate.
There is a reasonable argument for higher sample rates and more gradual high frequency roll-off than you get with 44.1k sample rates.
There is some very non-scientific speculation about interpolation and zero-crossings that seems blended with concern for dithering.
They start to invoke magic when claiming that something recorded with DSD then converted to PCM (and even further converted to MP3) will sound better than a direct conversion to PCM. If there is any truth to this then it would be more of a comment on the quality of modern ADC's rather than innate limitations on PCM or innate superiority of DSD. Again, though, their comments are general and anecdotal. The do cite a recent recording session they did where they recorded with a single mic to a DSD recorder and multiple mics to PCM, later mixed down. They note that the two methods aren't directly comparable and they also don't note which result they liked better (though one could assume) or offer characterizations of the two results. They do say they are planning another recording session taking the same analog signal and recording via DSD and PCM for a more direct comparison, but that hadn't been done yet (The video was made October 2018).
They attribute to DSD many of the characteristics and qualities people seem to attribute to listening to music via a Yggdrasil
Doug has great presentation skills and lots of experience. Well yes I have a lot of respect for experimentation. I’ve learned an incredible amount in life due to experimentation and using my own senses.
In regard to high frequency and strings, well we all have our own likes and dislikes. I know from personal experiences that both certain quality violins or poorly played violins can sound pretty bad to me up close. Doug use to record the Philadelphia Orchestra for them.
I know Doug is use to using a Radar system which has been regarded as one of the best PCM recording systems in pro-audio. He did say that there was not a huge difference between the two. I think the bigger point he was making is that down sampling/conversion to a lower quality format provides better results from DSD.
About the sampling rate of PCM not being about to do highs Justice. Well maybe the pro industry just does not have an equivalent filter to Mike’s......
If only Mike went Pro!
I thought they had offered files for comparison. I’m looking into that.
rkw
Headphoneus Supremus
Mike has been there, done that. He and Nelson Pass designed the GAIN 1 multibit ADC for Mobile Fidelity Sound Labs. Somewhere in this or Jason's thread, there is a list of CDs mastered with this system.
Last edited:
RickB
Headphoneus Supremus
Muddy Waters' Folk Singer comes to mind (MFSL).
bosiemoncrieff
Headphoneus Supremus
Just got the BluRay of the Barenboim 1991 Ring — should come any day now — as well as the Medici TV Max Lorenz Documentary "Wagner's Meistersinger, Hitler's Siegfried." That Thielemann/Domingo Parsifal is a weird one. The Goodall/Vickers is really sumptuous tho — I think the opera gets better the slower you conduct it. (Meistersinger, and to a degree Walküre, are the opposite. Tristan defies changes in tempi utterly—it subsumes them.)
audio philestine
100+ Head-Fier
I have a link in my archives to http://www.mofi.com/v/vspfiles/assets/images/goldcds.txt for a list of all MFSL gold CDs.
I also have archived a post from another audio forum that says the GAIN1 MFSL CD's start with UDCD 593 Muddy Waters Folk Singer and end with UDCD 734 Jethro Tull Songs From The Wood. I don't know if those are the 'official' start and end points, but I haven't found a more authoritative source yet.
Folk Singer is a perennial Audiophile favorite, and can be found for quite a premium on eBay. Hi-res versions can also be found on HDTracks, but I have no idea if they can be traced back to the MFSL sources.
ehayduk
New Head-Fier
- Joined
- Feb 3, 2011
- Posts
- 17
- Likes
- 27
Yes I understand that. I own quite a few MFR’s.
Great stuff. I’m talking about product today! AD/DA pro level
Last edited:
ehayduk
New Head-Fier
- Joined
- Feb 3, 2011
- Posts
- 17
- Likes
- 27
That recording is SCARY REAL!!!!!
I gave my copy away. The sad emotion that was captured was too intense for me.....
ehayduk
New Head-Fier
- Joined
- Feb 3, 2011
- Posts
- 17
- Likes
- 27
Oh so you own MoFi, very nice!! Thanks for yours, Mike’s and all others who have provided these great resources!!
I have been listening to lots of modern music of this genre recently but it is pretty hard to beat the old classics, what a great album!
audio philestine
100+ Head-Fier
Just to be clear, only the link was in my archives, but the list is on the mofi server, and I have no affiliation with mofi, other than owning a few of their CD releases, and hoping to ejoy more.
wout31
500+ Head-Fier
My Folk Singer album 192/24 is mastered and digitally remastered by Erick Labson. He's never done any mastering for MFSL as far as I was able to find.
Pietro Cozzi Tinin
Headphoneus Supremus
Whatever anyone tells you here...... DSD is really, really good.I also watched this, and while I also felt that some of the technical arguments came close but fell a bit short, there were a few take-aways
1) They are musicians (and a recording engineer) with well trained ears that tried a lot of equipment and compared and found a sound the liked, even if they can't explain the science in a way that I can buy into. They're hearing something, even if I can't explain it.
2) I interpreted their comments on the DSD recording sounding better when mixed down to mp3 as more a commentary on a short, simple, and refined recording chain. Very nice mics placed with thought around good musicians in a nice acoustic space, into nice mic pre-amps and direct hi-res digital recording. No mucking around with mixing or effects or questionable post processing techniques. Quality recording makes for a better end product, even after downconverting,
While DSD doesn't get any love around here, [personal opion/soapbox to follow; I'm sure not everyone will agree...] I get the impression that what they liked about the DSD recorder it is that it takes the internal oversampled representation (e.g. 1bit at 2.5+ whatever MHz) of the music internally in the ADC and records it directly, without attempting to downsample it back to PCM (24 bit, 96,192, whatever), and then feed it to an oversampling DAC that converts it back to the 1-bit MHz+ internally before making it back into music. That's a lot of extra conversion steps which historically have not been done well, so in theory eliminating those conversions should make for a better end product. But... most of us here have bought into the magic of baldr's expertise at doing PCM very well, rather than DSD's attempt to perform a technological run around. DSD assumes doing less is better than doing something poorly, but we've discovered doing something well is better still
Users who are viewing this thread
