When is Hi-Res not really Hi-Res?

[Bruce B]

Is this article an accurate indictment of the Stones hi-rez?

No it isn't... because...

1. These were DSD transfers and you would see digital high-frequency noise up to 50k, whereas his graph does not show it.

2. He has turned down the sensitivity of one graph to show that "it doesn't go above 14k". As you can see in the second graph... there is information all the way out to 41k

3, Thus, you are also missing the tape bias bump which his graph does not show.. My graph shows the tape bias at about 29k in both FFT and spectrogram

4 and most important. His other graph shows one of HIS recordings which he sells.

 

[Ron Party]

Good info, Bruce. Thanks.

 

[Orb]

Just to add,
as part of HifiNews June 2011 Hires investigation, Keith Howard purchased the "Through the Past Darkly" Rolling Stones album - 176k version.
His software showed that the low pass filter was applied around 30khz, however he felt that in this example it was not upsampled (is good when considering certain factors).

One thing that is of interest is he measures both peak and average spectrum; at 20khz to 30khz (relatively flat for that whole frequency range) the peak is -80db while average is around -90db.
As a reference point, roughly 2.5khz peak is -20db and average -30db.
The graph was showing "Honky Tonk Women".

Cheers
Orb

 

[Phelonious Ponk]

there are hundreds of plug-ins available for use in software like Audacity, Reaper, Pro-tools..etc. That's what a good ear is for. What plug-ins to use to make this track sound better. Maybe it needs more top end... Then you would need a parametric EQ plug-in to add a few dB of shelf EQ. If the track seems too dry, then you would need an ambience recovery plug-in.

No "one fits all" here. It just depends on what YOU think the track is lacking.

But at this point, you're re-mastering. Does the change in sound really have anything to do with the upsampling?

Tim

 

[Bruce B]

But at this point, you're re-mastering. Does the change in sound really have anything to do with the upsampling?

Tim

Upsampling a file does not make it sound better. It's still the same file. Now, if your DAC has a sweet spot at say 176.4, then upsampling is good. Upsampling is only done if the file is going to be processed further...

 

[Orb]

Yeah agree with you Bruce.
In Keith Howard's article he showed how several of those he downloaded ended up with slight flaws due to the upsampling.
In a way upsampled hi-res files are not really value for money as you do not get anymore information and have a chance of it being done incorrectly.

Cheers
Orb

 

[microstrip]

3, Thus, you are also missing the tape bias bump which his graph does not show.. My graph shows the tape bias at about 29k in both FFT and spectrogram

Bruce,
Can you guess what was the tape recorder used for this recording just looking at this spectrogram?
If you ever get sick of audiophiles and non-audiophiles, I am sure CSI will have a place for you.

 

[Bruce B]

Bruce,
Can you guess what was the tape recorder used for this recording just looking at this spectrogram?
If you ever get sick of audiophiles and non-audiophiles, I am sure CSI will have a place for you.

This was an Ampex/ATR-102

 

[Bruce B]

Here are some files I just ran across. The vendor says they are true 24/176.4 files. He sent some graphs that shows the spectrum.
I have drawn lines showing that 24-bit files "only" have a dynamic range of 144dB. Anything below that is noise. These are 24/88.2 files that have been upsampled to 32-bit/176.4
You can clearly see that any music content or noise goes silent at 41kHz

 

[Orb]

If this becomes more common Bruce,
I get the feeling this will become the next perceived snake oil, which will be a shame for those that are true hi-rez files.
Personally I feel this is a worrying trend, but glad the main audio magazines have started highlighting one needs to be careful.

Thanks
Orb

 

[Bruce B]

Well, the "Audacity Cowboys" are at it again over at the CA forum. HERE Go to end of thread.

wgscott and Teresa are saying the "Talking Heads - True Stories" album is not hi-rez. I see content out to 40k. This is what happens when you put tools into the hands of _______.

 

[amirm]

I was recently testing a bunch of things with Audacity and I was shocked how incorrect its frequency spectrum display is. If you don't know what it is doing, you can completely be misled. I find soundforge to also have issues although not as bad. The best I have found is Adobe Audition (Cooledit).

 

[microstrip]

All these issues are postponing my decision to go into Hi-Res.
Is there any reliable place where I can find information about which issues are or are not true Hi-Res?

 

[Orb]

HifiNews has started showing their analysis of hi-res music they review, they are using the structure-measurement as developed by Keith Howard several months ago.
In current magazine one album (think they reviewed 5 or 6 this month) was upsampled and they mentioned they will report it to the company and expect it to be removed or changed, btw the company was not HDTracks or other main hi-res download company.
Works well, one can hope other review publications will do something similar for the hi-rez review section.

Cheers
Orb

 

[wgscott]

Well, the "Audacity Cowboys" are at it again over at the CA forum. HERE Go to end of thread.

wgscott and Teresa are saying the "Talking Heads - True Stories" album is not hi-rez. I see content out to 40k. This is what happens when you put tools into the hands of _______.

Well, I am not sure how you want the blank filled in, although I can guess. However, it would extremely helpful to understand what I might be doing wrong.

To get some sort of handle on what I am observing, I have also included spectral plots from a positive control (Talking Heads, Remain in Light, Track 01, which looks very similar to what you posted), and a negative control (Lush Life, Track 04, which is from an album subsequently removed from HD track's catalogue). The first plot looks much more like the negative control to me than it does the positive control, and it looks more like the negative control than it does the plot that you posted. So you can imagine why a ______ might be concerned.


http://www.computeraudiophile.com/files/Screen Shot 2011-09-01 at 3.54.57 PM.png
Track 01 of True Stories

Positive Control:
http://www.computeraudiophile.com/files/Screen Shot 2011-09-01 at 4.13.54 PM.png
Track 01 of Remain in Light

Negative Control:
http://www.computeraudiophile.com/files/Screen Shot 2011-09-01 at 4.07.55 PM.png
Track 04 of Lush Life

 

[RBFC]

OK, this will be an interesting discussion, but let's keep it civil and informative. Any more personal attacks, etc. will result in administrative action being taken.

Lee

 

[fas42]

As a thought, if showing spectrum plots it would be a good idea to use the same sampling size and windowing function so that we're comparing apples with apples ...

Frank

 

[wgscott]

OK:

When I attach it, it shrinks the image, so here is the original:

 

[amirm]

First, welcome to the thread and forum .

Second, part of the explanation is easy. Look at your vertical scale. What do you see? It stops at -90db. Now look at Bruce's display. What is the lowest value? -120db. If you draw a line at -90db on Bruce's graph, you see that it also stops around the same spot as yours.

What I can't yet explain without access to the file is the brickwall effect you are showing in your display. That drop is not in Bruce's file.

 

[wgscott]

That's what I thought at first, too, but I think it is just that the normalization is different on the vertical axis (His ranges from -60dB to -150, and mine ranges from -18db to -90db). In any case, as you point out, the main feature is the abrupt change of slope at ~19-22kHz. That striking feature simply isn't present in the other plot.

The most reasonable explanation is that we are looking at data from two different files. You can see it looks much more like "Born under Punches," which is why I provided that as a positive control (and a guess that he might be confusing the two HDtracks track #1 Talking Heads files).