192kHz vs. 96kHz

[Randy Myers]

The few 96kHz albums I have bought the past couple days all sound amazing!

[BowiePop]

Originally Posted by Mike1998

I was just listening to the new Bowie collection this morning, you are correct, best I ever heard those tracks. Man Who Sold The World sounded phenomenal.

I'm totally agree with you!

[Randy Myers]

So I read a lot of people saying how incredible the recording of the Roger Water's album Amused to Death is. Being a huge Pink Floyd fan and always liking the music Roger was involved in the most I had to give it a try.

I decided to get the 192kHz version simply because of how well produced everyone said it is. First listening and it really is amazing. I doubt the 192 is noticeably better than the 96, but just could not resists because the quality rating on the album.

[jfrech]

Interesting thread. Wish I would have seen it sooner. My take is this all goes back to the source. If it's a analog tape...I want the highest resolution possible. 24/352 even. If it's a digital master with digital editing, I prefer to stay as close to the source resolution...

I have great sounding files at all resolutions. So I wonder if something else is going here besides just the sampling frequency of our playback files...

Speakers: Rockport Altair 2
Preamp: Robert Koda K10 Amp: Ayre MX-R
Cables: Transparent Opus MM2 Spkr, Opus Gen 5 & Magnum Opus Interconnect
Power Shunyata Triton v2/Typhon/DPC-6 v2, Anaconda Zitron cords (x6), Sigma HC (x1), Sigma Analog (x3)
Digital dCS Vivaldi DAC, Transport & Clock w/ Transparent Reference XL Digital cables
Phono: Grand Prix Monaco TT, Spiral Groove Centroid, Jan Allaerts MC2 Finish & Lyra Atlas, Nagra VPS/VFS/MPS
Racks: Grand Prix Audio Silverstone F1 (x2)

[jdcarlson]

I recently completed a three year process of digitizing all of my analog material to digital (to save space - my wife and I are moving into 1/3 the space we now have).
I made the original recordings at 192k/24bit - since that was the highest resolution I could afford at the time I started (I saw some 384K units advertised, but they were ten times my budget).
After making the original recordings, I electronically "down-sampled" them to 44.1/16 "Redbood" resolution to play back on the equipment on my system (McIntosh MS750).
On one particular recording (an old Jan August mono LP from my parents) the 44.1/16 was "unlistenable" due to the surface noise from the original LP (50 years old).
The analog original was very "listenable." The 192/24 digital copy was also "listenable" (but not quite as good as the analog original).
Perhaps the DAC in the playback machine was not the best (McIntosh MS750), which could have affected the difference. Perhaps a $15,000 DAC could have made the 44.1/16 "listenable" - but not the McIntosh equipment I had.

If we think about what happens when an analog signal is digitized, we can understand why. Think of a simple single1khz sine wave. At "Redbook" standards, this single sine wave would be cut into 44.1 "slices." At 192/24 this same sine wave would be cut into 192 "slices." It is obvious that 192 slices will be thinner, and thus have less vertical differences in height.

This is the acoustic equivalent of "pixilization" in digital photography. That is why digital photography seems to be racing to increase megapixels, with the latest Canon "35mm equivalent" digital camera having over 50 megapixels.

Perhaps a $15,000 DAC would have the circuitry to "smooth out" the "jaggies" but I know of no DAC affordable by mere mortals that can properly smooth out these "jaggies."

Based on my (admittedly limited) experience, higher resolution is always closer to analog (which is what our ears hear).

Thanks for listening,

Jim

[Cohibaman]

Originally Posted by jdcarlson

...If we think about what happens when an analog signal is digitized, we can understand why. Think of a simple single1khz sine wave. At "Redbook" standards, this single sine wave would be cut into 44.1 "slices." At 192/24 this same sine wave would be cut into 192 "slices." It is obvious that 192 slices will be thinner, and thus have less vertical differences in height.

Jim,

Very cool experiment!

What you say is mostly right, but one technical clarification; the 44.1 "slices" you refer to are really horizontal "samples" (the sampling rate) and pertain to the frequency response limits. The greater the number of slices, the higher the frequency response.

The vertical differences of these slices (samples) are what is referred to as bit depth or the volume level of each slice. A higher bit depth allows for more discrete volume "steps" or less differences in height between slices. Less differences in height means less less quantization error. Less quantization error means less quantization noise.

I'm not sure if the noise you heard after downsampling was caused by lowering the sample rate and bit depth in and of itself, but if it was, it was more likely from the lower bit depth.

I'm not an expert so I hope someone else can chime in on what else could cause the noise after downsampling.

[o0OBillO0o]

Have you tried other dither and truncating software?

Here is a comparison website: SRC Comparisons