Show Me Your HiRez

Mike Nomad · « **on:** 9 Feb 2011, 04:17 am »

Since we have a thread for "faux" HiRez, would it be useful (or at least interesting) to have a thread discussing the real deal? I finally got around to trying out HDtracks, and downloaded their 24/96 FLAC of Alison Krauss & Robert Plant "Raising Sand."

Out of curiosity, I ran the first track, "Rich Woman," through Audacity to see what the graph looks like. I was surprised by the bumps at ~28.75K & 45K...

Pez · « **Reply #1 on:** 9 Feb 2011, 05:11 am »

I guarantee those spikes are noise encoded in the track either from the recording itself or some sort of issue with the way the file was created. Not that it matters a whole lot given that only your dog could appreciate the flaw.

jamesg11 · « **Reply #2 on:** 9 Feb 2011, 07:51 am »

Mike, how does the krauss/plant 24/96 look in audacity with the amplitude -1 to +1 graph. The 16 bit cd was an unpleasant dynamic compression effort.

Mike Nomad · « **Reply #3 on:** 9 Feb 2011, 04:04 pm »

Quote from: Pez on 9 Feb 2011, 05:11 am

I guarantee those spikes are noise encoded in the track either from the recording itself or some sort of issue with the way the file was created. Not that it matters a whole lot given that only your dog could appreciate the flaw.

I didn't phrase my observation correctly. The one @ 45K, by itself, doesn't surprise me. White Papers I've read on DSD & LPCM both detailed how they put garbage in at half the sampling frequency or twice the understood top of human hearing. Wish I had an Audio Stunt Dog, to see if they would react.

The one @ ~28.75K is the one I don't understand.

@ James: Not a perfect haircut, the left has a few hairs at 3:22 & 3:40, while the right has one at :30. How does this compare to the 16-bit?

DSD_Mastering · « **Reply #4 on:** 10 Feb 2011, 11:46 pm »

It's called tape bias. It happens during the tape transfer.

Mike Nomad · « **Reply #5 on:** 11 Feb 2011, 04:18 pm »

Quote from: DSD_Mastering on 10 Feb 2011, 11:46 pm

It's called tape bias. It happens during the tape transfer.

Well, er, hmmm. Somewhere in the chain, they recorded on tape. Which would also explain the spike @ 45K. Or, are you saying the bias is the spike at ~28.75K? I didn't know tape bias was done below 40K.

Mike Nomad · « **Reply #6 on:** 14 Feb 2011, 07:49 pm »

From another HDtracks 24/96 FLAC download: Duke Ellington & Johnny Hodges' Side By Side, the track "Stompy Jones."

Never seen anything like this one before: 21K to ~22.7K is flat at approx. -82.5dB, followed by a slow rise to -81dB at ~31K, stays flat to 34K, then a slow fall to the floor (-90dB) at ~45.8K (with a not-surprising bump at 38.5K).

Any Ideas?

kyrill · « **Reply #7 on:** 16 Feb 2011, 04:46 pm »

Mike
Can you hear the difference between the normal CD and the hirez version of Raising Sand?

greetz
Kyrill

Mike Nomad · « **Reply #8 on:** 17 Feb 2011, 12:41 am »

Quote from: kyrill on 16 Feb 2011, 04:46 pm

Mike
Can you hear the difference between the normal CD and the hirez version of Raising Sand?

greetz
Kyrill

Kyrill,

My CD copy of Raising Sand is currently out on loan, so, I'm going from memory...

Yes, I hear a subtle difference. The HiRez copy has a touch more detail, and a lot more presence. If you are OK with the CD, I don't if you will feel like you'll get your money's worth from the HiRez download. I tried the download purely as an experiment, and don't regret it.

Mike

Napalm · « **Reply #9 on:** 17 Feb 2011, 02:59 am »

Quote from: Mike Nomad on 14 Feb 2011, 07:49 pm

From another HDtracks 24/96 FLAC download: Duke Ellington & Johnny Hodges' Side By Side, the track "Stompy Jones."

Never seen anything like this one before: 21K to ~22.7K is flat at approx. -82.5dB, followed by a slow rise to -81dB at ~31K, stays flat to 34K, then a slow fall to the floor (-90dB) at ~45.8K (with a not-surprising bump at 38.5K).

Any Ideas?

Few microphones go up to 45000 KHz.

Play it at half or quarter speed so you could hear what's there.

Nap.

HAL · « **Reply #10 on:** 17 Feb 2011, 03:12 am »

Quote from: Mike Nomad on 9 Feb 2011, 04:17 am

Since we have a thread for "faux" HiRez, would it be useful (or at least interesting) to have a thread discussing the real deal? I finally got around to trying out HDtracks, and downloaded their 24/96 FLAC of Alison Krauss & Robert Plant "Raising Sand."

Out of curiosity, I ran the first track, "Rich Woman," through Audacity to see what the graph looks like. I was surprised by the bumps at ~28.75K & 45K...

The upper spike looks to be at 44.1KHz. Possibly leakage from the CD A/D sample rate clock.

Mike Nomad · « **Reply #11 on:** 18 Feb 2011, 02:29 am »

Quote from: HAL on 17 Feb 2011, 03:12 am

The upper spike looks to be at 44.1KHz. Possibly leakage from the CD A/D sample rate clock.

You're right: I was off one hash on the grid. Your idea of the source noise is interesting: The track is a 24/96 FLAC download. I wonder if Chesky's guys are simply tweaking a CD master, and not sourcing the originals. That would explain what I'm seeing on the other graph...

@ Nap: I can half speed it, but I don't understand why: Graphing the half-speed will simply drop the top by half. If I do it again, the same thing happens. You suggestion leads in an interesting direction... halving the speed would put the top freq. around the middle of the flat in the low 20K range.

If this is what it looks like to me, bummer. The tracks still sound really good, and I'm glad I got them. However, it starts to look like these tracks aren't what they should be.

HAL · « **Reply #12 on:** 18 Feb 2011, 03:18 am »

I have seen that type of thing happen with both a master and backup copy are made simultaneously at the recording session. One recording being HiRez and the other CD rate. The CD sample clock some times bleeds over. The graph looks very smooth above 22KHz, so the master looks Hi-Rez. Those other spikes look like artifacts of the recording processes.

Napalm · « **Reply #13 on:** 18 Feb 2011, 03:26 am »

Quote from: Mike Nomad on 18 Feb 2011, 02:29 am

@ Nap: I can half speed it, but I don't understand why:

At half speed the content in the ultrasonic range becomes audible. So you could get an idea what it is. If it sounds like music in "slow motion" then it may be some useful signal. If it sounds like noise then it's noise.

You could even do more, like apply a high pass first so you eliminate the 0-20kHz signal then play in half/quarter speed so you could hear only the ultrasonic content.

Nap.

Mike Nomad · « **Reply #14 on:** 18 Feb 2011, 03:41 pm »

Quote from: Napalm on 18 Feb 2011, 03:26 am

At half speed the content in the ultrasonic range becomes audible. So you could get an idea what it is. If it sounds like music in "slow motion" then it may be some useful signal. If it sounds like noise then it's noise.

You could even do more, like apply a high pass first so you eliminate the 0-20kHz signal then play in half/quarter speed so you could hear only the ultrasonic content.

Nap.

Now I understand where you're going. I was (literally) looking at this from solely the graphical perspective. The high pass filter / low speed playback is a great idea. It'll be tomorrow when I can get around to it.

Mike Nomad · « **Reply #15 on:** 21 Feb 2011, 11:24 pm »

Well this is annoying: Not getting my high-pass filter in Audacity to work. Not sure how long this is going to take...

Mike Nomad · « **Reply #16 on:** 24 Feb 2011, 01:37 am »

Here's another graph: The track is S.K.J., from Milt Jackson & Wes Montgomery - Bags Meets Wes!

This is more like what I expected from the Ellington / Hodges record.

Mike Nomad · « **Reply #17 on:** 24 Feb 2011, 01:42 am »

Quote from: Mike Nomad on 14 Feb 2011, 07:49 pm

From another HDtracks 24/96 FLAC download: Duke Ellington & Johnny Hodges' Side By Side, the track "Stompy Jones."

Never seen anything like this one before: 21K to ~22.7K is flat at approx. -82.5dB, followed by a slow rise to -81dB at ~31K, stays flat to 34K, then a slow fall to the floor (-90dB) at ~45.8K (with a not-surprising bump at 38.5K).

Any Ideas?

After I bailed on Audacity and moved over to Reaper, success with high-pass filtering & altered playback speed. There is indeed music information above 23K, up to ~36K. After that it sounds like tape hiss, and once I get up to around 40K, silence.

DSD_Mastering · « **Reply #18 on:** 24 Feb 2011, 03:19 am »

Quote from: Mike Nomad on 18 Feb 2011, 02:29 am

I wonder if Chesky's guys are simply tweaking a CD master, and not sourcing the originals. That would explain what I'm seeing on the other graph...

Yeah... that's what we're doing.

Mike Nomad · « **Reply #19 on:** 24 Feb 2011, 04:40 pm »

Quote from: DSD_Mastering on 24 Feb 2011, 03:19 am

Yeah... that's what we're doing.

It would be a bummer if you were, and there are plenty of stories about companies not doing their best when passing something off as the best.