Hey Paul , Got the Reference SACD yesterday from Ocatave records , And I must admit i have 14 amps and a dozen set of speakers ,So i hooked my my Reference speakers then i put your SACD in and Wow i had my speakers Left on right and right speakers on left side can't believe i did that But your SACD told me quickly ! The phasing helped also ! Thanks again Paul Everytime i roll amps or speakers i will use your SACD
I'm not sure if my understanding is correct for PDM/DSD, but the way the density works is that for each 1 bit being transmitted, relative to previous bit it increase the voltage being generated on the DAC, and the reverse is true (for every 0 bit, the voltage drop up to the lowest). The sampling/clocking rate represent the resolution of the relative difference between bit on DSD and thus the data size is depend on the sampling rate 1:1 (e.g 2.6mhz equivalent to 2.6 million bits per second). This is the density or also known as rate of change in relation to previous data. On PCM, the way the data is being stored is separated into 2 dimension, where amplitude/voltage are represented by bit depth in Y axis and frequency are represented by sequenced of data in X axis. Thus each frequency being sampled are represented by a code with consistent word length, while the higher the frequency being recorded the higher the sampling rate required. So for a 16 bit 48 khz PCM will be 24x48000 of bits per second or about 1.1 millions bits per second. I still don't understand on the Delta Sigma relation of AD/DA converter though and I could be wrong in interpreting it. From the way the data being represented, DSD do seems mimic biological neural transfer as data are pulsed not on constant code but rather by the rate of change. CMIIW
Yes, you've understood this pretty well. In the density model of DSD the more "on" bits vs. the more "off" bits is what makes higher or lower voltage in cadence with the musical signal. The sample rate of DSD is fixed (as is the sample rate of PCM). They are very similar in some ways. Single rate DSD runs at what we refer to as 64fs. That is, the constant sample rate is 64 times faster than the standard of 44.1kHz. Simply multiply 64 x 44,100 and you get 2.82mHz (which is the sample rate of DSD). As you note, PCM uses fixed words (as opposed to the constant stream of DSD). Each "word" or collection of bits in PCM is defined by a fixed value. In the case of CD that value is 16.
Pretty good high level explaination Paul. I think those going to google might want to look up PWM (Pulse width modulation) as that might help explain the difference to PCM (Pulse coded Modulation). For those that don't know PDM is PWM with a specific clocking frequency.
@@misterbonzoid5623 Depends on the application where PWM is being used. Some have a specific refresh cycle time other uses my just depend on duty cycle alone.
I didn't hear you define 'DSD', and I found your explanation of how PDM works really confusing. You say at 4:20 'no one bits are happening, essentially, if there's no sound...' Silent passages are represented by streams of alternating 0s and 1s, louder ones by lengthening sequences of 1s alternating with lengthening sequences of 0s.
I'm developing an analogy using driving down the road. The road is very hilly, always going up or down at various heights above sea level (ASL). With PCM each sample distance is measured to find out how high it is ASL. An exact number of feet ASL is measured and that value stored as a sample. Then you walk a little further and measure how high ASL that spot is and store that as a sample. Then walk some more, measure ASL and store it... You can pull up that data file and immediately know at any point along the road how high that point is ASL. You could program a flying droid with those ASL data points and tell it to fly just above that and it should fly along the road blind just fine! Now take a paper towel tube. Lay on a spot in the road and look level straight ahead. Is the road immediately ahead of you going higher? YES or NO? If YES, a ONE bit is created. If NO, Zero Bit. Now crawl slightly to the next spot look through the horizontal tube. Is the road going up? YES or NO? That is all single bit cares about. Do I create a Zero or One? The question being with DSD is this new sample higher than the current status? YES or NO! A flying droid would be unable to start at some random point and know exactly how high off ASL it needed to be. But once it new where it was, it would know whether to go up or not with the next sample. That is all it knows or cares about. Up? YES or NO! So on playback if a sample is one, 1, the output is turned on and a pulse is generated. If Zero, 0, the output stays off for that cycle. The more consistently the output is turned on, the higher the output signal goes. The less often it is turned on, the output slowly drops.
I think many get lost when "1 bit" is mentioned. A PCM sample gives an amplitude representation (a number) at a specific fixed point in time (defined by the sample rate). PDM just encodes amplitude over time so it's really not "1 bit" per se but a stream of bits and how they are spaced given a known clock frequency.
@@mbfishing769 PDM is truly 1 bit. Each sample is merely a YES/ NO, One/Zero, 1/ 0 based on current input compared to existing output. Either YES the voltage has increased or NO it has not. It does not care where the voltage is overall. It does not even care if the voltage has stayed the same or gone down. Just "UP? YES/NO"! Just "do I turn on the output YES/ NO". If playback is randomly started in PCM, the first sample will say exactly what voltage to start at. If playback is randomly started in PDM, the first sample says nothing about the original signal voltage level at that starting point. The first sample just says "Turn on output? YES/ NO". Same with the next.... After a few samples it will catch up to the original signal.
@@glenncurry3041 For sure, we are just saying the same thing in different ways. My "PDM just encodes amplitude over time" is the same as you said "After a few samples it will catch up to the original signal". For your playback example, one PCM sample = exact voltage where as the PDM will require many bits over time to reach the same voltage. I was just trying to point out that many get lost when saying "1 bit".
@@mbfishing769 I was just trying to get away from the concept of PDM "encodes amplitude over time" because PDM doesn't care about what the amplitude's actual value is. It encodes the positive only changes over time. I think that is where a lot of confusion comes in. It was some of mine. I kept wanting to know how a single bit would know a specific amplitude at a specific spot. I finally understood it don't give a ...! Just "Pulse On? Yes/No!" But with today's use of Sigma Delta instead of ladder DACs as they do not generate a specific voltage, they are one bit, many PCM DACs "will require many bits over time to reach the same voltage". My first exposure to PDM was Kenwood FM receivers in the '70's came out with PDM detectors that merely took the heterodyned frequency modulated signal, rectified and filtered it!
@@glenncurry3041 It can be a troubling concept to explain. Many times I've had luck with mechanical thinkers using PWM analogies (PDM without a clock signal). E.G. With a 2 ms pulse a servo motor sits in the middle of its range, a 1ms pulse width causes the servo to rotate 90° left of center and a 3ms pulse moves the servo 90° right of center. Of course all pulse lengths between go to their relative positions, those are the absolutes. Now imagine that servo moving a speaker cone, servo goes left and it pulls the cone in, servo goes right and pushed the cone out. Works for some poeple ...
I thought I had a good understanding of DSD coding, until I heard your explanation! And then I went to Wikipedia, and it turns out I understood it just fine. Sorry Paul but your best advice in this video was just like my Mom used to say to me all the time...... "Go look it up". My favorite book growing up was my set of encyclopedias. Remember those??
Paul keeps banging on about the very high sample rates used for DSD compared to PCM. The implication is that the higher the sample rate the better the audio quality. This is a total misunderstanding of how digital recording works and the two formats cannot be compared in this way. Unlike PCM, DSD produces stair steps in the output waveform and very high sample rates are required to make this distortion inaudible. Well recorded PCM and DSD music will be audibly indistinguishable. DSD was obsolete the day it was invented.
“If you can't explain it to a 6-year-old, you don't understand it yourself,” Albert Einstein. Also using the word 'bit' to mean both 'digital value - 0 or 1', and 'amount' is confusing, as at 4:03.
By "density", I am assuming that our host is referring to the percentage of how many of the 11,000,000 bits are present each second? Because there is no such thing as a computer storing code more densely (the closest thing would be a zip file or some other form of lossless compression -- but all that does is do character substitution to store fewer bytes -- it still uses the same sectors and allocation units on the storage device in exactly the same way for every file). But all files (compressed formats and uncompressed formats) are stored exactly the same way on each storage device (all determined on how you had your operating system format the drive -- or how the factory formatted the drive -- assuming you did not re-format the drive). Regardless of how analog DSD bits look when an app displays them on a graph, DSD is 100% digital. It is 0% analog. Granted, it sounds very much like analog (I will take our host's word for it). But it is still 100% digital.
It's about the 'density' of longer sequences of 1s alternating with longer sequences of zeros increasing with volume I believe. The data rate is constant. Unless I've misunderstood. en.wikipedia.org/wiki/Super_Audio_CD Also, very much niche, as the Digital Audio Workstation software everyone uses can't use native DSD/PDM or has that changed?
The more dense or more 1s are closer to each other,(i.e 10111011101 versus 00010010101) the louder the signal is being sent to the speakers. Do that milions of times per second and you get reproduction of a maximum of 512 times higher sample rate than CD(on dsd512, the highest dsd version available)
How hard is it to modernize that old Sonoma Windows XP software to run on Windows 11 and/or Mac OS? What programming language is that Sonoma system written in?
I wish it were easy to do but it's not. The entire program is written in embedded C and relies upon a collection of custom FPGAs. The UI is written in something else that I don't know about, and could likely be redone with great expense and time, but no need. Pyramix has already done the work and offers a modern system based on DSD up to 4X. We are simply switching from Sonoma to Pyramix.
@@Paulmcgowanpsaudio Yes, sounds like a fairly big engineering project to recover that system. Perhaps Pyramix should add DXD+ (705.6kHz or 768kHz) and Octave Records could be the one customer that is pushing Pyramix for next generation fidelity on both DSD & PCM?
Hey Paul, you keep repeating the same thing over and over. The real question is with two DSD sound tracks that you want to mix together, how you do this without either doing it with your analog mixing board or converting to another digital format?
Well, on some subjects I have said the "same thing" in a different way for dozens of time. It's what helps people come to an understanding. If I say it one way then a small handful get it, but a large section doesn't. On the second go-round a few more, etc. DSD cannot be level changed or mixed. It must first be converted to either multibit or analog. To make it multibit it requires a phase perfect digital low pass filter (to do it right), which is what our Zephiir filter's all about.
@@octaverecordsanddsdstudios1285 Is Zephiir some kind of proprietary or a trademarked name that just maintains the ambiguity of what your may be doing just to keep the mystique going?
@@octaverecordsanddsdstudios1285 I'm just worried about over-selling DSD as it doesn't plug into what creates earnings for musicians. I just went to the John Mayer concert yesterday (in Hollywood) and learned (again) how distorted live concerts are. His SACDs from 20 years ago sound better than more recent music from him. How do you attract someone like him or anyone else making great music who is already popular? You gotta convince those guys by connecting to the master quality streaming services from Apple, Amazon and anyone else starting to get into 192kHz 24 bits or potentially better (you can think DSD is better, but ignore that please), cause no money = no honey for this business.
@@chrisharper2658 Hi Chris. It is a proprietary filter though someday we'd love to share it with the world. Basically, in order to convert DSD to PCM all one needs to do is run it through a low pass filter. In most cases, that's what happens in a DAC. If you want to mix or control level in the DSD stream you must first either convert it to analog or PCM. In either case, you run DSD through a low pass filter to do that. If you're going analog you use an analog low pass filter. If you're staying in the digital domain you run it through a digital low pass filter. Zephiir is unique in that it is phase perfect-something neither an analog LP is nor your traditional digital LP.