I agree. It is still PCM.
I just find it funny how the DSD advocates try to label it "DSD wide" to hide the fact that the "pure" DSD has to be converted to PCM and back.
Hey, bavmike, julf, etc. Do you know how many data format conversions are there between speaker terminals, and pleasure neuron networks in human limbic system and pattern recognition glutamate neuron networks in auditory cortex? In which format is music stored in neuronal networks? Do the neurons work in pulse density mode? Are brain waves the clock?
Against this I can only cite some amplifier porn: http://www.diyaudio.com/forums/class-d/307102-prototyping.html#post5065905
What most PCM advocates forget is all modern ADC's use SDM technology. So when you have a PCM track it was converted to PCM from
DSD in the studio DAW already. When you have a DSD album sourced from a DSD direct recording, or analog tape, if you have a pure DSD DAC this lossy conversion never needs to take place. And to apply DSP to it, doing it at it's native rate is the best possible way. It's the same way the best DAW packages edit DSD, only done in realtime during playback. But takes incredible resources to accomplish. Far more than possible on weak Sharc chips.
All modern ADCs use sigma-delta modulation, yes. That is not the same as DSD. Most of them convert directly to PCM as part of the (essential) filtering process. Very few ADCs talk DSD between the ADC and the DAW.
In any case, as you acknowledge, the signal has to be converted to PCM for any processing in the DAW (and then back, if you want DSD output).
Regardless of what's done in the DAW, the biggest benefit of DSD is it allows you to bypass the resource constrained SRC/SDM faculties in all modern ASIC DAC chips. However when you use the grade of SRC/SDM I'm using in software with my system, the source format is no longer of much importance. The grade of the recording is what matters most.
Seems you were looking to start a DSD vs PCM debate. Most of those debates are pointless because what matters is always ignored. When you start comparing high grade software based SRC/SDM to ASIC based SRC/SDM, what matters becomes very clear. Most DAC's today use ASIC chips mainly designed for PCM processing. AKM's have the advantage of a DSD bypass mode that allows bypassing the resource constrained SRC/SDM section. Sabre chips don't have this bypass option, but when you feed them DSD, you automatically bypass the SRC section. It's this section that's responsible for the famous "harsh Sabre sound" that many folks don't like. This is the main reason folks like the sound of DSD better. It's simply due to their DAC's being poor at processing PCM. ASIC DAC chips are built to a price point, and also designed for low heat dissipation, and small size. So they are extremely limited in the DSP section. This is why many DAC manufacturers who even use the Sabre, use Sharc, or FPGA based SRC before the chip. But still that's only a half assed fix. But good enough for the price point of these units. As most are sold through a distributor/dealer model they need to keep production costs extremely low, in order to keep profits high, and deliver the goods at a reasonable price to the end user.
Last edited:
No. I was just pointing out that "DSD wide" is a misleading term (on purpose) as it it actually is PCM.
I wouldn't really call it misleading. Because anyone who uses the term clearly explains what it means. It's important to differentiate this from simply calling it conversion to PCM. Because that usually involves lossy downsampling. This lossy downsampling is what all current DSP chip based systems do because they don't have the resources available to process the DSP on the DSD at it's native rate.
Digital Scam for Dummies? No way to represent more data=entropy with less bits, except redundancy compression. Data loss occurs when attempt is made to sample delta-sigma pulse trains, which by definition are free-running oscillation between a positive and a negative error, exactly in the instant the error is exceeded. It would be like sampling the PWM from an UcD at fixed rate, and using that bitstream to reconstruct back the audio.
You can explain that to an audiophile until you're blue in the face. At the end of the day what sounds better to them is what they will like better. For me personally, a competent system is one that can process both DSD and PCM albums equally well. PCM obviously much more important as that's what 99.9% of all source material is available in.
What does it happen when the outputs of two complex circuit networks, producing exactly opposite signals, are fed to a summing amplifier, and gain of all elements is increased to the point that quantum mechanics start to make the output bounce? This is neurology of superstition. There are chemical reactions in the brain where factors can be made to cancel out such that result depends on quantum mechanics, like flipping a coin. Reality perception can be configured for superstitious or rational mode. Problem with superstitious mode is that what is picked up (or synchronized to) by the neuronal "antennas" is quantum vibration from another individuals or things, so this is essentially a defense mechanism when learning is jammed as due to abuse or rewarded deceiving. Evolutively superstitious mode is used for syncing with targets such as food or other individuals, particularly when the organisms are in low resource situations (can't afford iron for glutamate for rational mode). Audiophiles perceive interesting things but mostly not as sound pressure waves.
Last edited:
I would. Because it isn't DSD, it is PCM.
They do? Oh, OK... 🙂
Despite the fact that it is conversion to PCM?
I'm more interested in the results rather than terminology. This is the problem with explaining how anything is actually done. The best way is to seal it all up like a black box, and leave it up to listening tests only to determine if a good job was done. Otherwise you end up with both under and over educated, real and armchair engineers deciding how the product sounds without even listening to it.
Create an emotional support group with more rational certainty than the audiophiles and even the most superstitious will feel tempted to buy from time to time. But their lack of certainty will add load on the certainty of the support group, so frequent training in rational-reality-connecting activities is required, as those where it is required to bet for optimum reliability solutions to measurable survival problems, for example living space creation and refurbishment of aged ones knowing where it fails first.
I know. But a common, accepted and precise terminology makes for better communication.