Question from a listener:
I know this topic is a bit controversial, but I was wondering what NativeDSD's take is on DSD 64. In one of the audio forums I frequent, there is a DAC manufacturer there who also participates there. Their take on DSD was that not an ideal format to use because of the computational power it needs to filter out the high-frequency noise, the fact that the signal is shaped by noise makes it mathematically approximated, and the market for true native DSD formats is very small. They ended up creating a DSD-capable DAC for DSD 64 playback (DSD 128 required them to redesign their architecture), but compared their normal PCM-playback DACs, they don't sell very many at all.
I've read elsewhere that DSD 128 and above don't have some of the problems that DSD 64 has, but native DSD 128 recordings are incredibly rare (I did purchase A Far Cry's Dreams and Prayers album in DSD 128) on top of the already pretty niche native DSD albums.
Answer by Tom Caulfield - mastering engineer
No controversy. Just two university profs designing a poor mathematical paper Delta-Sigma modulator, then bashing it. It's very old news from the dawn of DSD, when Sony/Philips hoped to replace the CD with a new licensable optical format; SACD.
DSD and PCM are the two coding formats used today in the recording and delivery of audio content. DSD (Direct Stream Digital) is a marketing term for a specific variant of Pulse Density Modulation (PDM), and was developed to achieve a more accurate method of representing an analog signal than is available with PCM in the digital domain. Its original intent by Sony was for archiving tape during the 90's. After further development by both Sony and Philips, it became the delivery format for the then emerging Super Audio CD (SACD).
DSD has the advantage of sounding more "analog like" (ie, accurate) than PCM, because it is actually an analog format storable on a digital media. While in principle it can operate at any bit rate, at the time of first commercial adoption, 64fs (2.82 MHz) was the practical limit. That bit rate was high enough to shift the inherent un-correlated random noise component above the 20KHz audio band of interest. However, while there is no interaction between this high frequency noise (shifted noise) and the audio signal, because the noise is un-correlated, and well below -100dB in the audio band, it can affect downstream equipment. That being marginally stable amplifiers principally. This is where it gets its bad rap about the effects of high frequency noise. However, this is actually is quite rare, and more an urban legend.
To your question about bit rates, every time you double the bit rate (64fs > 128fs > 256fs etc.) of DSD, you shift the un-correlated noise envelope one octave higher. That is, with 64fs where the noise was -120dB at 20KHz and -60dB at 50KHz, at 128fs, it will become -120dB at 40KHz, and -60db at 100KHz. At 256fs those -120dB and -60dB frequency numbers double again. The unfiltered overall noise energy remains the same, regardless of bit rate, its beginning presence just gets shifted proportionally higher in frequency as the bit rate is increased. This makes it easier to passively filter in the audio stages after the D/A conversion.
So what does this do to the audio signal fidelity and accuracy within the audio band? Nothing, in my experience. But since the perception of sound is such a personal and subjective experience, it's easy to suggest it must be better. That sells new stuff!
Nativedsd.com is in the music delivery business. Our emphasis is in the offering the closest to the origin, the native, obtainable music content. That means edited masters, the last stage of recorded music production, and prior to authoring to a delivery media process. The best way for you to answer your question(s) is to experience the results yourself.