Topic: DSD over PCM

[H₂O]

http://www.positive-feedback.com/Issue59/dsd2.htm


Interesting read


I have noticed the market is currently flooded with home built 32bit 384Khz/ DSD128 USB DAC's (search ebay)


This is neat in the sense that newer hardware maybe able to support DSD over a single Coax SPDIF style connector instead of traditional SDIF-3 which requires 1 BNC connector for each channel and 1 for clock (3 for stereo, 9 for eight channels, etc)


Also it may make it more trivial to interface DSD signals with standard CPU's allowing development of professional DSD recorders at a lower cost

Some additional information:
http://wiki.jriver.com/index.php/DSD_Format

[H₂O]

Here is a up and coming Korg series (an older version is out abroad as well):


http://www.korg.co.jp/Product/Audio/DS-DAC-100/

[F.O.Bean]

I have been HEAVILY contemplating get 2x D100 or save up a lil more and get an SD 744!?!? I like the option of DSD, but I don't see myself using it at festies anyway. I would just run 24/44.1 at festies and save DSD for single shows! Whereas I would just use the 744 the 9 months of the year that aren't festies. I'm not sure which way to go, buy now I'm leaning towards 2x D100s!!!

[page]

Also it may make it more trivial to interface DSD signals with standard CPU's allowing development of professional DSD recorders at a lower cost

I think it knocks down reduces a barrier to native editing. (which I think is one of the barriers to adoption of it) Doing a DSD>PCM>DSD production sort of defeats the point IMHO. The sooner that middle PCM step for editing can be eliminated within a large number of DAWs, the better off the format will be.

[Gutbucket]

I don't think it does anything to knock down the barrier to native DSD editing, but agree that problem is the fatal flaw of the format.  This is simply a transmission work around, and DSD editing the far more fundamental problem.  This disguises the DSD data as PCM to allow transmission across USB though the Apple or MS OS.  As far as editing is concerned the data is still DSD. PCM-based editors and tools can't do anything with it without conversion.

I don't mean to rain on any parades, but personally I don't get the DSD thing.  It seems to me to be an unecessary complication, difficult to manipulate for creators and editors while offering little if any appreciable benefit over hi-rez PCM to end-listeners, the vast majority of whom find it unecessarily confusing.  In the final analysis I don't think it's a postive thing for the greater goal of getting high quality music to the people, but rather a conduit for nitch marketing and another way for audiophiles to feel superior and insulate themselves.  Most of the motivation I see is from those positioning themselves to take commercial advantage of that, rather than compelling reasons of audibility.

If it made a significant difference to record as DSD then convert to PCM for editing and playback I could accept that as a good reason for it over recording directly to high rez PCM, but I'm unconvinced that it does.

In my experience, attention to better recording and production is what really makes the difference, and more channels easily wins over super-resolution formats. 

[edit- but having said all that, this development makes moving DSD data around much easier, and that's cool.)

[page]

[edit- but having said all that, this development makes moving DSD data around much easier, and that's cool.)

Thats basically it; it makes it much easier to move stuff around. It may not knock out that barrier, but it does reduce it.

[macdaddy]

I still think dsd serves a purpose when transferring master analog tapes, when any mastering is done in the analog domain. But for our purposes, without true editing in the dsd domain, the format's use is limited at best...

[DigiGal]

I have been HEAVILY contemplating get 2x D100 or save up a lil more and get an SD 744!?!? I like the option of DSD, but I don't see myself using it at festies anyway. I would just run 24/44.1 at festies and save DSD for single shows! Whereas I would just use the 744 the 9 months of the year that aren't festies. I'm not sure which way to go, buy now I'm leaning towards 2x D100s!!!

Bean,

Take a look at the new SD 663 - http://www.sounddevices.com/products/633/

[H₂O]

Personally I do very little editing as I am more of a purest then one that likes to Produce the raw audio in order to compensate for room issues or equipment limitations - if I pull a tape that's base heavy or light on the vocals I chaulk it up for what it is - although I find this is quite rare I produce an unlistenable tape

there is DSD editing software out there but it is a bit expensive at around $800 or so

The key here is that the DSD realm is starting to pick up a lot of steam:
  - some record labels now selling DSD downloads
  - high end consumer receivers supporting DSD playback such as pioneer and Onkyo
  - high end AD and DA companies are starting to include DSD input/output as standard such as Mytek and Benchmark
  - many other smaller DAC releases such as Korg, and Teac

It only a matter of time when editing capabilities are added to common editors such as Wavlab etc and possibly free software (the foobar support is open source)

ASIO 2.1/2.2 supports DSD so part of the equation is already there

[F.O.Bean]

I have been HEAVILY contemplating get 2x D100 or save up a lil more and get an SD 744!?!? I like the option of DSD, but I don't see myself using it at festies anyway. I would just run 24/44.1 at festies and save DSD for single shows! Whereas I would just use the 744 the 9 months of the year that aren't festies. I'm not sure which way to go, buy now I'm leaning towards 2x D100s!!!

Bean,

Take a look at the new SD 663 - http://www.sounddevices.com/products/633/

Ive LUSTED over that puppy for awhile now Im just not sure if I want to stay in the 24bit realm for another couple years, or try something different and start recording DSD. Im not in a huge hurry to produce DSD to the masses, thats what 24bit/16bit is for. I just want that HQ Super-Res SOURCE so that I CAN enjoy it when/if it FINALLY spreads to the masses!

I think if I were to stay in the 24bit realm, I would HIGHLY CONSIDER the SD 663 4 channels for my Schoeps, and 2 extra channels for SBD patches, that I RARELY ever take And Id still keep the M10s for festies anyway And I agree with H20, Super-Res HQ sources arent going to change anything if what you recorded is shit anyway. I also agree with him that I RARELY have a recording that isnt playable. You have to get that good location or good venue to pump out the right stuff to get a good recording, and DSD isnt going to change that

Im just one confused taper

[F.O.Bean]

I have been HEAVILY contemplating get 2x D100 or save up a lil more and get an SD 744!?!? I like the option of DSD, but I don't see myself using it at festies anyway. I would just run 24/44.1 at festies and save DSD for single shows! Whereas I would just use the 744 the 9 months of the year that aren't festies. I'm not sure which way to go, buy now I'm leaning towards 2x D100s!!!

Bean,

Take a look at the new SD 663 - http://www.sounddevices.com/products/633/

Also, the ONLY thing Im not vibing with of the 663, is it ONLY has 3 frickin preamps Didnt SD learn from the 744 If it had that extra preamp, id probably DEF grab one, but that means id still have to use a preamp for channels 3/4 If I lay down that much $$$, I dont want to lug around external preamps and having to stress over powering the external pres

[stevetoney]

So, from quality perspective, if you're a live music taper and you're not currently recording at 24/192, why would you be concerned about potential future upgradability or potential sonic improvements of DSD?  Asked another way, if you're not currently taking advantage of the resolution capabilities available to you with your current rig, why would you be interested in upgrading to DSD?

I'll answer my own question from my perspective...please tell me if my logic is flawed. 

Relative to a higly controlled and quiet studio environment, I record sh**ty sounding shows in sh**ty sounding venues.  Since I have sufficient resolution to do anything in post that I need to do to my 24 bit recordings without loss of sonic quality, nothing I record is gonna sound any better if I record it with higher resolution anyway. 

To me, it's kinda like having a super high quality one million megapixel camera body, but you're using your camera with low quality lenses to make kodak moment snapshots.  Beyond a certain point, concerning yourself with the megapixels might be nice, but in practical application, it makes no practical sense relative to the end game.

[H₂O]

Remember that when your audience taping your also taping the room not just the band - I find my DSD recordings sound more open and have more of the room ambience then my 24 96khz PCM recordings

The DSD recordings give you a more of being there feeling

[page]

Remember that when your audience taping your also taping the room not just the band - I find my DSD recordings sound more open and have more of the room ambience then my 24 96khz PCM recordings

The DSD recordings give you a more of being there feeling

I can respect that opinion because of your goals, and if you're getting results that you prefer via DSD, thats great.

I'm working in the opposite direction in trying to eliminate the room as much as possible when I record, and because the focus is different, my requirements and factors for consideration are as well. (and it's that change in factors that causes DSD to falter for me).

When you say that DAWs will be adding DSD support, will their plugs be downsampling before processing, and then upsampling back to DSD?  A lot of VST (PCM) plugs will upsample (because they need to), but often they don't tell you that.  It would be a *lot* easier to write a single DSD-->PCM-->DSD routine and drop that into all of your existing DSP code than it would to recode all of your routines to native DSD (not even sure how possible that is).  It would also be a lot smarter to run a single downsample in front of the entire processing chain and a single upsample at the end, because that way you'd only have one set of filters rather than 10 or whatever, which would preserve both quality and CPU cycles.  Which gets back to my point about the major use of DSD being to reserve decimation for an offline process where resources are not limited.

I did some casual reading yesterday while I had some time, and what I'm at a loss on is the noise introduced by doing the DSD>PCM>DSD conversion vs just oversampling in PCM. Resampling adds low level noise, sure, but is the format conversion an equal/lesser/greater noise introduction, and is that noise more bothersome than just the PCM resampling noise. That I don't have an answer to.

[Gutbucket]

Remember that when your audience taping your also taping the room not just the band - I find my DSD recordings sound more open and have more of the room ambience then my 24 96khz PCM recordings

The DSD recordings give you a more of being there feeling

I find that very interesting, since it is one of the top-level recording goals for me.  Besides the technical analysis Jon suggests (which I fully agree is more illuminating of actual differences in output, if not of subjective interpretations of differences), it would also be interesting to audibly compare recordings made simultaneously from the DSD and PCM outputs of the same converter to listen for differences.  H20, If you ever get a chance to do that I'd be interested to hear your observations on the aspects you’ve noted.  Whether any perceived differences are due to differences in converters, low level distortions, psychoacoustics, listener bias or whatever, the subjective impressions would be interesting, specifically in light of their contrast to, or correlation with those reached by technical analysis.

..my DSD recordings sound more open and have more of the room ambience then my 24 96khz PCM recordings

The DSD recordings give you a more of being there feeling

I mentioned above that’s a top goal for me, and I completely understand if you choose not to explore this route (you certainly would not be alone), but if that immersive ‘feeling of being there’ is important to you and you find the pursuit of even subtle improvements in conveying it worthwhile, you might consider recording for more than two channel playback.  That’s been my primary motivator in moving to recording for 3, 4 or 5 transmission channels.  When done well, the contrast with 2-channel stereo is not subtle, it is immediately apparent and downright tangible.  Although there are a number of other playback aspects which also benefit, the “gives you more of a feeling of being there” thing is one of the most shockingly obvious ones both in my own listening experience and from the comments I hear most often from friends for whom I get to properly play these recordings.  At this point that experience is not an easy one to share, but it conveys (future-proofs if you like) the ‘you are there’ aspect of the event to a far greater extent than relatively subtle differences from recording to higher resolutions.  It is at least as dramatic and obvious a difference as going from mono to 2-channel stereo, and arguably more so.  Forget about commercial surround music releases, I'm talking about live music recordings made more in the 'purest techniques' sense like we do around here. 

To me the return on investment in recording for additional playback channels is so much higher than that from improved conversion or increased recording resolution above a certain threshold (for me that threshold is 24/48PCM with a decent current-day ADC, for others it may be 24/96, or 16/44.1 or whatever) that the decision of which route to take in pursuit of that goal has been an easy one.  I acknowledge it’s a heavy investment- not only in recording and playback gear, and in recording expertise, but perhaps most importantly in a greatly reduced ability to share the experience widely, at least currently and into the near-future.  Yet for me it's worth all that.