How did we get to 192?

July 17, 2022
 by Paul McGowan

When playing high-resolution PCM files the defacto industry standard seems to be 192kHz, 24 bit. Right? I mean, given a choice between the next candidate, 176.4kHz, we automatically choose 192kHz.

Why would we do this? Perhaps human nature. We assume higher is better.

But, in fact, there's reasonable evidence that in many cases, 176kHz is preferred. Much depends on the original recording process.

At the dawn of digital recording, the pro machines were all based on the standard of 48kHz (and later its multiples: of 96kHz and 192kHz). To produce useable digital releases on the (then) new CD format, everything had to be rejiggered through a complex compromise that resulted in 44.1kHz for CDs.

Life at the time would have been a lot simpler if CDs had just adopted 48kHz as the standard, but, alas that was not to be.

Today, some recording engineers still mindlessly choose 192kHz as their high watermark to record at, then downsample to the uneven result of 44.1kHz (which is always a bit of a compromise).

If one were to think about it just a little, we'd be asking for releases and recordings to be standardized at 44.1kHz and its multiples: 88.2kHz, 176.4kHz, 352kHz, etc.

It' probably doesn't matter much anymore and it's certainly nothing to lose sleep over.

But if I have a choice it's always 176kHz.

Subscribe to Paul's Posts

17 comments on “How did we get to 192?”

  1. Yes I think Ted also mentioned this some time ago.

    The other question for me is, what your new DSD256 format means for PSA DACs (I remember generally for single bit DACs), which according to Ted have a better sounding sweet spot between DSD128 and DSD256 (means DSD128 sounding better than DSD256 or higher on those DACs ). This due to quantization noise getting lower with higher bit rates, but jitter getting higher then (DSD64 being the best for lowest jitter but the worst for quantization noise vice versa with DSD512).

    Did you try if DSD256 makes sense on a PSA DAC anyway, because other factors rule out the above mentioned suboptimal noise behavior over DSD128?

    I see a bit of an analogy with today’s post when asking “how did Paul get to 256?” 😉

  2. Interesting topic. It begs the question then: why 2 of the 3 Octave Records PCM downloads come in multiples of 48KHz …Instead of 88.2 and 176.4 … The conspiracy theories abound… were they chosen mindlessly or for some nefarious reason. 😐 😀

    I guess numbers sell….
    Like you I prefer even multiples….
    BTW …Sleep deprivation hasn’t been an issue worrying about about sample rates and bit depth on this end.

  3. 48KHz makes for a more reasonable guard band than 44.1, which needs a "brick wall" filter with a lot of phase distortion. It is also more compatible with 24fps movies and 30fps video. For early engineers making one-off machines it made the most sense.

    Sony, being a heavily top down operation, had an executive mandate to fit entire Beethoven symphonies on one disc, and squeezed the sampling rate to 44.1 to get 74 minutes play time. Sony, again, picked 64x44.1 = 2822.4KHz so SACDs would be compatible with 44.1 circuit components!

    I am not sure that dithering integer conversions produces better musical results than fractional. I would bet that for decimating DSD masters it is a nominal difference between 44.1 and 48. Are there direct comparisons examples online? Any that use different dither algorithms to compare?

    Since DSD is about the same file size as 24/96, 176 or 192 seem wasteful. I have almost 4TB of DSD, so my collection would not fit on a standard FAT disk at 24/176.

  4. The two base frequencies is maddening. It is too bad, really. Requiring two base frequencies in DACs, or the use digital synthesis to derive the correct frequencies. More beating clocks of disparate frequencies in DACs are more expensive, and can interfere with each other, requiring extensive engineering to deal with. Also, we have the problem as noted of non-integer conversions. Now, with more powerful processors available, and very powerful computers, non-integer conversions can be done without problems, but it would be easier to just be able to avoid them altogether.
    Additionally, the DDS clock generators have become much much better these days. They are so good now that a good engineer can actually use a reference clock of 10 MHz, coupled with a good DDS, and actually achieve lower phase noise than by using individual XOs of the correct frequency (given the advantage of the lower phase noise of the 10 MHz clock in the first place due to its lower frequency).

  5. I just bought a bunch of downloads of of Craft Records Lost Recording’s series. They are all sampled at 176.4kHz/24 Bit. They sound fine to me.

  6. Every D to A converter I've encountered sounds different at different sample rates and not always better at 192. That said, my testing has shown that 96 always sounds better converted to 44.1 than 88.2 or 176.4 do assuming that a modern sample rate converter is being employed. Back in the '90s even multiples sounded better because of limitations of the available computers.

    1. Guenter Pauler, head of of Stockfisch records and Pauler Acoustics (who experiments a lot, makes fantastic recordings on various formats, but whom I don’t necessarily count as a pope of digital detail knowledge) as far as I remember favors 24/88.2 as the best format and sees disadvantages in higher sampling rates.

      One gets as many opinions as people asked, it seems.

  7. I make DSD recordings of some of my vinyl records. I do not convert them to anything else. I simply play them back on my PS Audio DS DAC. I think they sound almost identical to the vinyl.

    1. I don’t doubt that generally or for you explicitly, but I think it’s always very easy to experience test scenarios where comparisons on that level of subtleness make no difference. I would even say that applies for most scenarios. Sample rate differences, Flac/AIF differences or such you mentioned. Most don’t hear it.

      For an objective statement, as always, it’s just important if there are scenarios where it makes a difference, as then there IS a difference…just not for everyone and every scenario. Nobody who hears no difference in anything is wrong, he just can’t generalize his experience 😉

      1. jazznut, sorry for not getting back to you yesterday. The TASCAM hi-rez recorder that I use records in all the variants of PCM up to 24 bits / 192 kS.p.s. and DSD. When I first got the unit ( circa 2010 ) I made recordings of the same track using PCM at increasing sample rates and also DSD. I then played them back on the TASCAM which has both record and playback capabilities. Thus, this was a true apples to apples comparison.

        From this I can tell you that the largest improvement in SQ comes when you simply double the sampling rate from 44.1 kS.p.s. to 88.2 kS.p.s. Overall I found DSD sounded the best. I find ( as some people do ) that there is a sound to PCM that DSD does not have. DSD sounds almost identical to the analog.

        1. Thanks! That’s roughly what I expected from what the label owners say…you’re probably one of the few others who could meaningfully experience this yourself.

          I can just listen to what’s offered and that on a DAC that anyway converts everything (which is great, but difficult for comparisons)

  8. Dan Lavery has written that the ideal rate mathematically is 60 with signal processing done at multiples of that. Unfortunately, digital audio only works financially with mass produced components so we are limited to multiples of 44.1 and 48 or, thankfully, DSD.

  9. acuvox - agreed.
    They wanted a reasonably decent storage system to get decent sound from a 72 minute disc that would play on the widest variables in manufactured CD playback offerings. The compromises were struck, and the ndustry followed, history was made.

    4th generation players (around ‘85) became pretty good. Mine played in my system until recently, only because I wanted SACD.

    1. They really wanted a digital disk that could be pressed in a vinyl plant. They merged Sony's digital recording via video tape system with Philips' failed video disk system that employed vinyl record presses. The 44.1 sample rate was what the Sony technology and its existing chips used. The 5 inch disk was a shipping cost saving measure.

Leave a Reply

© 2022 PS Audio, Inc.

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