Test CD for measurements of CD Players

[Luca Sorrentino]

What model? How old is it?

Hitachi DAP-XP2E. I suppose from the 2nd half of 90s. My measure was done on the line output with jogging buffer deactivated.

Cheers,
Luca

 

[NTTY]

Hello everyone,

When performing quality checks of the test CD files, I discovered two errors which remained hidden so far.

Two Intersample-overs files have unwanted digital distortion added to them.

I’ve requested @AnalogSteph to confirm the issue and he worked hard over night to recreate them with a new process of his, and I tested successfully this morning.

The two files in question are ISO 5512.50Hz and 7750Hz. So, don’t use them, they don’t show what we want.
@pkane : You might want to have a look, maybe I wrongly used Multitone from the beginning (look at H3 distortion).

The 11025Hz works well. Stick to this one when measuring Intersample-overs, it can be trusted.

I apologize for the inconvenience and I will update ASAP the Test CD and Excel.

To everyone who requested the files: no need to write to me, I’ll resend a new link for you to download the new version. In the meantime, just don’t use these two files.

————
Flo

 

[pkane]

Hello everyone,

When performing quality checks of the test CD files, I discovered two errors which remained hidden so far.

Two Intersample-overs files have unwanted digital distortion added to them.

I’ve requested @AnalogSteph to confirm the issue and he worked hard over night to recreate them with a new process of his, and I tested successfully this morning.

The two files in question are ISO 5512.50Hz and 7750Hz. So, don’t use them, they don’t show what we want.
@pkane : You might want to have a look, maybe I wrongly used Multitone from the beginning (look at H3 distortion).

The 11025Hz works well. Stick to this one when measuring Intersample-overs, it can be trusted.

I apologize for the inconvenience and I will update ASAP the Test CD and Excel.

To everyone who requested the files: no need to write to me, I’ll resend a new link for you to download the new version. In the meantime, just don’t use these two files.

————
Flo

Apologies, I at some point added the logic to clip the test signal, which obviously interferes with ISO signals that are greater than 0dBFS

What sample rate, bit count and signal length would you want? I can upload the test signal files for all three ISO tests without the extra harmonics.

EDIT: posted a version of Multitone that generates ISO signals without attempting to clip them to 0dBFS:

https://audiosciencereview.com/foru...loopback-analyzer-software.27844/post-2124793

 

[GWolfman]

Much appreciated!

 

[NTTY]

Apologies, I at some point added the logic to clip the test signal, which obviously interferes with ISO signals that are greater than 0dBFS

What sample rate, bit count and signal length would you want? I can upload the test signal files for all three ISO tests without the extra harmonics.

EDIT: posted a version of Multitone that generates ISO signals without attempting to clip them to 0dBFS:

https://audiosciencereview.com/foru...loopback-analyzer-software.27844/post-2124793

This was fast! Thank you.

We’re into 16bits/44.1kHz for the test CD. Thanks again for the great support!

 

[Luca Sorrentino]

A newbie question: how a signal can be greater than 0dB? They're not ±32767 full scale 0dB value?

 

[staticV3]

A newbie question: how a signal can be greater than 0dB?

Inter sample overs

 

[Luca Sorrentino]

Inter sample overs

I understand that an interpolation filter can generate samples higher than 0dB, and so the sense of this test, I just don't understand how can I generate and save a file with 16bit of resolution greater than 0dB.

 

[staticV3]

I understand that an interpolation filter can generate samples higher than 0dB, and so the sense of this test, I just don't understand how can I generate and save a file with 16bit of resolution greater than 0dB.

Most signal generators like Multitone Analyzer, REW, etc allow you to generate and save test tones with dBTP above 0dBFS.

These just place the sample points appropriately, to be interpolated above full scale:

 

[Luca Sorrentino]

Most signal generators like Multitone Analyzer, REW, etc allow you to generate and save test tones with dBTP above 0dBFS.

These just place the sample points appropriately, to be interpolated above full scale:
View attachment 402728

Perfect, I catch it, thank you very much. Just playing with the phase probably.

 

[dualazmak]

I usually use old but useful "Super Audio Check CD 48DG3 by CBS/Sony (1983)".

I shared the color spectrum (by Adobe Audition 3.0.1) and analysis data by MusicScope 2.1.0 of all the tracks of this "Audio Check CD 48DG3" in my post #651 on my project thread.

The CD booklet is in Japanese, and I myself translated it into English as attached herewith in PDF.

I also shared five video clips of "transient-sound" tracks within this "Audio Check CD 48DG3" played and analyzed by MusicScope 2.1.0 in my post here #760.

If you would be seriously interested in using and/or testing intact 44.1 kHz 16 bit PCM tracks of this "Audio Check CD 48DG3", please PM me.

 

[danadam]

Just playing with the phase probably.

In case of Fs/4 (so 11025 when 44100 sampling frequency) a square wave is equivalent of that and imo the easiest to generate.

 

[ElJaimito]

Well done @NTTY !

When your post first appeared I read the title and thought of posting a nice photo of a Stanley tape measure but then I started to read and it looked like you're doing a different kind of measurement. I figured it wouldn't be friendly so I desisted.

And to anyone else who uses this guide, don't forget to donate to REW.

Yes, probably overdue for another donation!

 

[VientoB]

This is awesome. I hope more people post their CD player measurements on here thanks to this!

 

[NTTY]

Hello Everyone,
I created a new version of the test CD (7.1.01) with:

• Updated intersample-over test tones to fix clipling
• Updated low level test tones files for better precision
• Added square test tones (100Hz and 1002.27Hz)
• 3DC level check now with 997Hz instead of 1kHz

Link to new version already sent to everyone who asked for the previous one. If I forgot someone, PM me.

Thank you!
--------
Flo

 

[NTTY]

Hello everyone, and the first player to benefit from a full review with the latest version of the test CD is my favorite Onkyo C-733.
It’s not the best CD Player ever but it’s good enough to be a challenge to beat.
I will update my other reviews when time allows.

 

[MC_RME]

Please note that the current iteration is still a "prototype", but it's already very advanced. Few highlights of what makes this test CD unique:

• All 0dBFS test tones are in reality @-0.008690235dBFS to account for 2's complement asymmetry

If I understood/read the reason for this symmetrical file correctly then it is that you got distortion before. But that must have been a fault in the generator. The asymmetrical full scale 16 bit signal would not show distortion at all - the difference positive/negative is too small in the 16 bit domain.

Adding a proper full scale test signal makes sense to test that this valid (!) signal is processed correctly. It also might be useful for other tests. For example it is possible that units with level meters use only that -32768 part to light up the Over LED (instead of the other method of counting consecutive three +-32767 samples). So IMHO it makes sense to add this full scale signal.

 

[AnalogSteph]

Oh, this'll be fun to generate. Basically you need a sine of amplitude 0.9999847412109375 with a DC offset of +1/2 LSB = 0.0000152587890625 (= -96.329598612473982468396446311838 dB). Let's just say Audacity wasn't exactly designed with such exotic needs in mind.

I did succeed in generating this eventually, but alas 999.91 Hz had suddenly partially lost its self-dithering properties (?!). 997 and 1000.03 looked better though. 1000.7 had very low-level traces of even-order HD, so on a hunch I checked the actual levels of my exported files, and what do you know, the +32768 value does not seem to be used at all and Audacity clips everything to +/-32767 = +/-0.999969. Grrr.

This one may be out of reach with amateur means right now.

 

[danadam]

All 0dBFS test tones are in reality @-0.008690235dBFS to account for 2's complement asymmetry

From the amount and from the linked post it seems like it's accounting more for shaped dither than for asymmetry.

the +32768 value does not seem to be used at all

+32768? The range in 16-bit is from -32768 to +32767.

This one may be out of reach with amateur means right now.

]$ sox -D -r44.1k -n -b16 sin_999.91.flac synth 30 sin 999.91 -0.00152587890625 norm

(result in attachment)

]$ ffmpeg -i sin_999.91.flac -af astats -f null /dev/null
...
DC offset: -0.000009
Min level: -32768.000000
Max level: 32767.000000

The first sample with value 32767 is at position 11, the first sample with value -32768 is at position 430. Still looks self-dithered I think:

 

[NTTY]

From the amount and from the linked post it seems like it's accounting more for shaped dither than for asymmetry.

Correct, only when exporting with shaped noise after creating a test tone of amplitude 1:

+32768? The range in 16-bit is from -32768 to +32767.

]$ sox -D -r44.1k -n -b16 sin_999.91.flac synth 30 sin 999.91 -0.00152587890625 norm

(result in attachment)

]$ ffmpeg -i sin_999.91.flac -af astats -f null /dev/null
...
DC offset: -0.000009
Min level: -32768.000000
Max level: 32767.000000

The first sample with value 32767 is at position 11, the first sample with value -32768 is at position 430. Still looks self-dithered I think:
View attachment 403620

Fantastic! Here FFT 128k, 32 averages and after conversion to WAV with Audacity (not sure of the potential negative impact of that and more so after reading the post of @AnalogSteph above), and comparison with the test tone at amplitude 1 directly created with Audacity (no dither of course):

Can you create directly in WAV format? I can't use a flac file from REW of Multitone.