DSD vs FLAC

August 15, 2021
 by Paul McGowan

27 comments on “DSD vs FLAC”

  1. I have purchased several Octave CD sets. I have a PSA Nuwave DAC which I bought to experience DSD & it’s my only DSD decoder. I have all of my CDs ripped to FLAC but the Octave recordings only include WAV files in addition to DSD causing me to have to convert WAV-FLAC to play on other systems. Why doesn’t PSA- Octave include FLAC files? Would you consider in future releases? Or am I missing something?

  2. Great topic.
    As almost a side-note…
    I’m running a streamer that plays native DSD files beautifully.
    It also “upsamples” PCM to DSD for playback very well, but the subtle sound difference is still there.

    What’s the next Octave Record release?

  3. Anyone know of any good tracks recorded in DSD which are on Qobuz/Tidal? Obviously they would be streamed in PCM but we would all appreciate the benefits?

  4. Paul, with respect this is becoming very confusing for people so please, please can you elaborate?

    My understanding is you don’t actually capture the initial incoming analogue audio signal in PCM as most A/D convertors these days capture PDM in the front end before it is converted ‘in the box’ to PCM so what’s happening is you’re ‘siphoning off’ the PDM as 1bit PDM (DSD). As it is in itself an analogue capture (even though it is 1’s and 0’s bits) it is closest to the incoming analogue audio signal so it makes sense that it will sound ‘better’ than the PCM it otherwise gets converted as a secondary process.

    It does still make arguable sense to convert PCM files into DSD on playback regardless of how it was recorded if the DAC is Sigma Delta based as these DAC’s will have more ‘ease’ in accepting a DSD stream for conversion.

    1. It is true most modern ADC/DAC use a multi-bit version that’s sort of like PDM but is in actuality PCM. They definitely use SDMs (Sigma Delta Modulators) and therein can lie many problems. Look in the last or second to last issue of Copper Magazine where Richard Murrison gives a much better explanation than I.

      The A/Ds we are building for Octave studios are running at 128fs and then rely upon the TI ADC’s internal SDM to produce the 2X DSD PDM stream. Once we have that we will use Richard Murison’s Zero phase decimator’s low pass function to then produce a perfect 352kHz PCM version of the original 128fs DSD capture. This then allows us to edit and mix in a souped up DAW without any sonic loss. The final 2-channel master then runs through the same A/D and we get the final 2-channel master in 2X DSD which we can then do what we want with it.

      This will be an absolute first and cannot wait to share the results with folks.

      1. This looks marvellous Paul. I hope it all works as expected. Just a question.

        You write “This then allows us to edit and mix in a souped up DAW without any sonic loss. The final 2-channel master then runs through the same A/D and we get the final 2-channel master in 2X DSD…”.

        Does this mean that the DAW outputs an analog signal which is fed back into the A/D?

        1. Yes, that is currently the plan. Ultimately we’d never go to analog at all, but what’s limiting us is access to a SDM that is better than what’s available internally to our DACs. That’s the holy grail and one we don’t yet have.

          We are working with a team of Greek mathematicians to design one but…I think somewhere there’s got to be a joke about a group of Greek mathematicians but can’t think of one. There are several on the market but so far we’ve not been impressed.

  5. I have one of the Octave DSD recordings purchased as a download and through my Aries Femto Streamer and Burson Conductor 3XP DAC it sounds excellent. I haven’t listened with my G2 and Chord DAC yet. Does it sound better than all my FLAC files ? Absolutely not. Some are better, many are worse. There is no doubt in my mind that production and mastering has a lot more influence on sound quality than Codec, sample rate or format.

    What is nice is having kit which lets me play pretty much whatever I choose ( apart from MQA which I could care less about ). Variety, as lazy writers might say, is the spice of life.

  6. I have been doing some poking around on portable media players on the internet in genirol.
    Some will play DSD while others will play both wave, and mp3.
    I believe that I saw one out there, that can decode and play WMA.
    But I haven’t found one that can play audio file types like APE and FLAC.
    Your iPhone and iPad, can play ACC and AC3 files as well.
    But they can also be made to play DSD too.
    But you have to download apps like Foobar2000 and plus the SACD input component in order to get any iOS device to play DSD audio files.
    Perhaps maybe Foobar2000 can play FLAC files, I just haven’t tried it with that one yet.
    But to my ears, DSD sounds way better then FLAC does.

    1. Hi John.
      I think the vast majority of Digital Audio Players can play FLAC and usually also DSD. Certainly my Shanling M0 can, and it’s a budget player. Unless I am confused by what you mean when you say portable media player.
      FLAC is pretty much the portable player standard Codec these days, WMA having fallen out of favour a bit due to larger file sizes and DSD still suffering from a lack of content.
      Obviously overall MP3 is still ubiquitous but of less interest to audiophiles.
      I would be surprised if you can find many players who don’t do FLAC.

      1. Good afternoon DIgriz!
        I don’t like MP3 files myself.
        To me, MP3 files, sounds like they’re watered down.
        I guess it’s because of the compression scheme that’s used in it.
        That portable player you spoke of, is one of the players I saw in the line up.
        But what you said about WMA files, I didn’t know that about them.
        Thanks man for the information on that!
        But to better explain to you about what I mean about portable media players, a small player that you can either clip on to your belt, or put in your pocket.
        They usually come with a pare of headphones or a pare of earbuds.
        I call them portable media players, because most of them, you have to buy batteries for them.
        But the high rez media players, usually have rechargible battery packs built in to them.
        By plugging them in to a USB port on your computer to put files on them, will charge them up for you, while you’re putting files on them.

  7. Oops. Sorry, John, I just realised I was having a senior moment and talking rubbish. Particularly getting WMA mixed up with WAV. WMA are compressed and smaller than FLAC but poorer quality, WAV are uncompressed but bigger than FLAC. It is WAV which I meant to say are becoming less popular IMHO.
    Sorry for the confusion.
    Shanling do two small players which happily do FLAC and are of a size to be used for sports etc. I can highly recommend the M0. I have had mine for a couple of years and it’s great. There are a number of other Chinese brands now doing small high quality players in the same segment of the market

  8. I appreciate Octave records experimentation with DSD. However, I second Dlgriz that “there is no doubt in my mind that production and mastering has a lot more influence on sound quality than Codec, sample rate or format.”

    I’ve purchased 3 Octave recordings. None of the versions sounded better than the vast majority of the 1,000+ admittedly curated-for-sound PCM/FLAC CD rips in my collection, or much of what is in my playlists from Amazon Music HD.

    My rig is pretty revealing. It consists of an almost-all PS Audio high end system: P15 regenerator, a DSD DAC (Senior), a BHK preamp and 250 amp, and Focal 1038be as well as Klipsch Forte IV speakers. All my power cables, interconnects and speaker cables are very high grade without going broke Audioquest. Speaker placement, which can have a great impact on SQ, was carefully done.

    Almost all of my recorded music, either ripped from CD or streamed from Amazon, exceeds the sound quality of all three of the Octave recordings I purchased. Play in either vinyl or stream the Mercury Living Presence recordings of Shostakovich’s 5th Symphony or Tchaikovsky’s 1812 Overture, and with any decently resolving system I trust you’ll easily hear the difference that production and mastering can make, even in the early days of stereo back in the 1950’s and 1960’s.

    There are many other examples, all of which point to production and mastering being most important. I wish Octave well. In time, the promise may fulfill the hype. As of now, to my ears and in my acoustically quite decent space, expectations of Octave productions simply are not being met.

    I value PS Audio tremendously. The gear excels. I also admire Paul’s dream of cradle to grave control over the musical listening experience. However, I fear overreach. A good recording takes a great deal of effort, as is the creation of a new speaker line. I would rather see PS Audio speakers finally come into production than recordings that are no better than others that can be obtained elsewhere.

    It is not my business to run but I very much enjoy music through PS Audio gear and hope that speakers and recordings that truly excel will see the light of day. I also hope that in this pursuit I do not see PS Audio six feet under.

  9. Very enlightening to learn more about DSD and PCM. Thank you Paul.

    A followup question on your comment about about FLAC. FLAC is, as you described it, a compression codec. However, and a reason I converted my former AIFF library to FLAC several years ago, is because FLAC conversion or CD ripping can be set to uncompressed (see https://drive.google.com/file/d/1OWOl5K5G8gKSUiGShiMY2KcLzo-nx0Yj/view?usp=sharing) at least with using dBpoweramp.

    So for PCM, is there any difference between FLAC uncompressed and AIFF?

  10. Thank you Paul for another good video .. I love the fact that a company such as PSA is using DSD recoding.
    From what I got from this video is DSD is best when it is captured using DSD, however converting PCM to DSD is almost the same as PCM or no change, and converting DSD to PCM is also the same or no change. PCM is used for editing/manipulating a file.

    My question: is all SACD players convert DSD to PCM to the DAC and then output the file as an analog signal? or DSD out to DAC to analog?

    1. Most legitimate SACD players have DACs built in that accept directly the DSD signal and output it in analog. That keeps copyrights in place and gives one the best sound possible. Our SACD player, the PST, is one of the few that directly outputs the DSD signal to external DACs capable of rendering it.

  11. We’ve visited this topic several times, and it’s all getting quite confusing.

    In a reply here, Paul confirms that the standard Octave Records workflow is to record 2X DSD, then convert to 24-bit 352 kHz PCM (also known as DXD) for processing, and back to DSD for delivery. That makes perfect sense, providing you have a DAW that can work at the 352 kHz sample rate.

    Digital recording is bit perfect, so we should assume that it doesn’t matter at which point in a processing chain the recording takes place. What I mean is that DSD → Recorder → PCM should be interchangeable with DSD → PCM → Recorder.

    24-bit 352 kHz PCM is very different from 16-bit 44.1 kHz, so it is entirely possible that complaints about the “sound” of PCM will not apply to DXD.

    The original Sony DSD initiative was based on recording the 1-bit delta-sigma format available at a test point in a PCM ADC. In this case, the PCM output was always transcoded from PDM, because that was the architecture of the ADC. Does this mean that the PCM output sounded nearly as good as DSD?

    It would be astonishing it any transcoding operation improved sound quality; all PDM/PCM transcoding is bound to be slightly lossy, even of the loss is so small that it cannot be heard.

    Many modern high resolution ADCs and DACs use multi-bit delta-sigma. Does these have the good attributes of PDM, or are they basically just PCM? What ADC does Octave Records use to record 2X DSD?

    1. Thanks, Mark and just to be clear, that is not what we are currently doing at Octave Records. What I described is where we’re going.

      Currently we, like the other audiophile labels who also work in DSD, record the original tracks in DSD directly out of the A/D converter. In our case because we use the Sonoma system, we’re only 1X DSD (64fs). Others using the Pyramix system, record in up to 4x DSD.

      Once recorded and edited, it is then ready for mixing.

      Mixing is done exclusively in the analog domain. So everything is run though high end D/A converters, and then passed on to a big analog mixing console. Once mixed and then mastered, it is then run through an A/D converter again and output as DSD.

      Because DSD is so close to analog, this process works and sounds great and there’s very little lost.

      Our new plan is to use the Pyramix system to record in 2X DSD (with the option to go to 4X if our ears tell us that’s best), then, using Murrison’s proprietary Zephir low pass filter in post production, convert to DXD for mixing in a DAW that we’ve designed to handle those higher frequencies. The final mix is then either run through an SDM for conversion back to DSD (if we can find or design one we like) or a final stereo D/A, A/D.

      The advantage of our newer design system will be to mix in the digital domain without degradation. The analog mixing process, while really good, can be bettered.

  12. Hi Paul, thanks for responding to my initial request for more info.

    So to clarify, at the moment you capture in DSD64 (Sonoma), convert to and mix in analogue then ‘recapture’ back to DSD 64? (similar to Cookie Marenco who captures in DSD256 dislikes DXD but also finds the Sonoma system too limited so also mixes in analogue and recaptures back to DSD256.)

    And your plan is capture in DSD128 in Pyramix, convert to 352.8 PCM (DXD) mix, then convert back to DSD128?

    If this is correct then i don’t understand – folks have been recording in DSD256 for many years now and converting to DXD for creating the final edited master. From there they can produce a DSD conversion for say SACD or download and any variety of PCM for CD/Blu-ray, download etc, or you can quite often now just download the DXD edit master itself. Converting DSD to DXD or high sample rate PCM whatever you want to call it would appear to be negligible in terms of losing fidelity but surely the ‘holy grail’ is to keep DSD as pure DSD without any conversion to PCM (difficult I know but folks are doing it).

    A question I asked was “why don’t folks just record directly into DXD when they know they are going to end up converting anyway? i.e what’s the point of capturing in DSD? It was explained to me that capturing into DSD and then after recording converting to DXD in post production can produce a more ‘robust’ DXD conversion as opposed to just recording on the fly directly into DXD (a question of CPU power). It also means they have the original DSD takes to keep for future use when they can possibly be mixed without conversion to PCM which is what these folks would prefer to do but can’t because of the workflow they are following. It would seem at any rate that capturing into DSD in the first place does give some advantage to overall fidelity as opposed to going straight into PCM.

  13. Thanks, Johndeas and yes, you’re basically correct. It’s what Cookie does as well as Robert Frederich etc. They choose DSD 256 (which we too might) and then convert to DXD. Couple of problems with this.

    The standard conversion to DXD via Pyramix sounds (to our ears) not so good (I was going to write dreadful but in a kind mood today). This is one of the reasons we didn’t go with Pyramix in the first place. Secondly, their SDM isn’t much better. So, what they get right is the recording process. What we’ve done is (through the Bit Perfect folks) to find a lossless means of getting to DXD (let’s just call it PCM because DXD is a kind word for folks alergic to PCM). Had it not been for the BP Zephir zero phase filter we would never have chosen Pyramix. The problem with the Zephir filter is it cannot be used in real time (it’s a look forward scheme).

    One could theoretically record directly in PCM but running 32 channels of 24 bit words at 11MHz is daunting from any number of aspects. Easiest to run single bit at that frequency then take your time to low pass it properly to something usable like 352kHz/24 bit.

    1. This is fascinating stuff Paul! You and your collaborators are obviously doing some thorough research into this and I’m excited for you. May you discover the mother lode in your digital explorations. 🙂

      PS Perhaps you could consider using a quantum computer in order to solve the ‘look forward’ problem? 😉

Leave a Reply

Stop by for a tour:
Mon-Fri, 8:30am-5pm MST

4865 Sterling Dr.
Boulder, CO 80301
1-800-PSAUDIO

Join the hi-fi family

Stop by for a tour:
4865 Sterling Dr.
Boulder, CO 80301

Join the hi-fi family

linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram