My daughter came home last evening for the holiday weekend. She’s a senior interactive designer at Modcloth.com in San Francisco and has become quite the expert on interface design and organizing artists and coders to design, implement and deploy cutting edge web stuff. She also got to me the new Waldrep puppy at dinner last night. Charlie, a 4-month old Border Collie, has won our hearts and keeps my wife and I on our toes…lots of runs, hikes and time at the park with the Frisbee.
Kari and I got up early this morning to run yet another Will Rogers 5 K through the neighborhood (it’s becoming a father/daughter thing). This was the 37th annual event and we triumphed once again…although, Kari beat me by mere inches at the final sprint. Her comment to her mother, “I wasn’t going to let some old guy beat me at the final stretch.” Thanks for the endorsement Kari (I did beat hear last year!). So the morning is behind me and I’m spending an hour or two at the studio catching up on emails and putting out the daily blog post.
Here’s something that crossed my desktop today.
What is up with the guys at Light Harmonic? They’ve broken through $800K with their Indiegogo campaign for yet another portable high-resolution audio player. That great news for them but I continue to be confused about how their supporters evaluate the merits of a device that seems somewhat redundant…and unnecessary. And the graphics and video on the campaign page are almost as preposterous as the ones on SuperHiRez.com. Check out this gem:
Figure 1 – The graphic showing “The Highest Resolution Audio Available” from the Geek Wave campaign.
I guess the creators of the illustration can do anything they want when it comes to skewing the truth but it pains me to see such gross misinformation. The x-axis includes same two completely different specifications. It’s simply not possible to list PCM sampling rates (with 24-bits per sample) in ascending order and then tack on DSD 64 and DSD 128 at the highest end of the chart (which are 1-bit systems). PCM uses multibit words and DSD is obstensably 1-bit, although many systems are actually multibit because of the limitations of working in 1-bit.
Here’s a note I received from John Siau, my go to electrical engineer and digital audio guru (head of Benchmark Digital), when I asked him about the use of Delta Sigma conversion in PCM devices:
“Every Delta Sigma converter uses noise shaping but not at 1 bit. Most converters are 4-bit Delta Sigma designs and the means that the noise shaping has an 18 dB advantage over DSD. It also means that dither can be applied. 1-bit systems are always under dithered. This fact is unavoidable. 1-bit systems can only be partially dithered.
It is important to note that in a DSD system the noise-shaping process happens every time the 1-bit signal is processed. Mixing, editing, and mastering can add many cascaded noise-shaping processes. This cascading degrades the DSD audio quality quickly. Some systems avoid this by processing in DSD-wide (which is a PCM system).
The decimation and interpolation processes in PCM systems are much less destructive than the 1-bit noise shaping processes that are required in DSD.
The questions are:
Would you rather have 1-bit partially dithered noise shaping or 4-bit fully dithered noise shaping?
Do you really want to do the noise shaping every time you mix, edit, or master?
24-bit PCM is a robust system that allows many cascaded processing steps with minimal degradation. DSD degrades quickly when processed.”
I’ve taken the liberty to rework the graph to bring the hyperbole back to reality. Here’s a better representation of the “potential” fidelity of the various sample rates/formats.
Figure 2 – This chart shows the reality of PCM and DSD fidelity. The Geek Wave does include the DSD formats but they aren’t on the upper end of the fidelity spectrum. [Click to enlarge]
So the reality behind the “spin” put out by LH is they are building yet another dedicated portable player the includes DSD up to 128 (equivalent to about 96/24…if you forget about the noise) and 384 kHz/24-bits, which is so far beyond what’s needed that it’s pure fantasy. Engineers aren’t making 384 kHz/32-bits recordings!
They would do better to focus on the analog electronics of the outputs, maximizing battery life and getting the sound right rather than pushing bigger numbers in our faces. Although, I guess they would response by pointing at the $800,000 dollars they’ve raised in this initiative. I stand corrected.