James-F wrote:

Consumer level audio (sound cards, MP3 players, etc..) will output 0.447V peak, so 40mV will be about -20db below peak. This is […]
Show full quote

stamasd wrote:

Working on getting some measurements in... I have a few problems with the test setup.
1. Many audio cards give me an output signal that's too low for some reason for my oscilloscope to process correctly... it sees the signal as being only 40-50mV for some reason, despite the sound being loud and clear when I commect headphones instead.

Consumer level audio (sound cards, MP3 players, etc..) will output 0.447V peak, so 40mV will be about -20db below peak.
This is very normal.

http://www.sengpielaudio.com/calculator-gainloss.htm
https://en.wikipedia.org/wiki/Line_level#Line_out

Maybe, but my laptop output's is about 10x that (I did some preliminary tests with dosbox before moving to real sound cards).

With several real cards, the signal is just enough to display (barely) but not large enough for the DSP to apply the fast Fourier to it.

Little known fact is Dosbox output is compressed and clipping at all times, not at all like a real sound card.
You should add "mixer master 40" command to your dosbox autoexec.bat so it will match the level of a typical dos sound card (my YMF719) and not clip at all times.
This is NOT equivalent to lowering the Windows mixer, because dosbox (SDL) already compressed/clipped the sound, so it should be lowered at the emulated sound card before dosbox output.

Setting dosbox mixer master to 40-50 will actually let games play at their original dynamic range and let it some space.
The game will reach -1.0db or so from time to time but only at the loudest moments, instead of reaching 0.0dbFS even at the quietest sounds.

I will do more precise experiment later this day to match the volume of DOSBOX and YMF719 outputs exactly.
This should give a dosbox mixer Master level which is most accurate.

That makes dosbox sound nicer, but doesn't help me bring the sound volume from my cards near something the FFT can use. I may have to use an amplifier - which will introduce its own noise and distortion. Also doesn't help Adlib Tracker II play nice with real OPL3 chips.

You should float the ground of your oscilloscope, I think it's a ground problem.
If there is a buzz as loud as the signal, the scope will not pick the signal up.
50mV should be easily readable, even 50uV if the signal is clean.

As for dosbox volume, it is obviously a bug.
I shall report it.

James-F wrote:

Hear what?

Lower pitch of 289 in the comparision sample on that article. What makes the sound noticably worse.

The sample song in wikipedia is absolutely the worst you can compare by because it has down-bending notes

If it has music examples of real output made on those chips, then it's meaningful.

all created purely in ones mind

To make such assertions (even subjectively) about pedia's examples you need preliminary to do their ABX comparision. I'm sure you did not. For my mind the difference I've heard is enough. Maybe later, when I'll get more basis to be paranoid about my mind, I'll play in ABX game with them.

Ace wrote:

I notice the drop in sound pitch only with specific instruments.

But it is, anyway. You hear it. I hear it. While James-F theorizes it's impossible. 😀 Even after he've heard the difference on pedia's example, he still thinks it's just the delusion of his mind.

James-F wrote:

You should float the ground of your oscilloscope, I think it's a ground problem. If there is a buzz as loud as the signal, the s […]
Show full quote

You should float the ground of your oscilloscope, I think it's a ground problem.
If there is a buzz as loud as the signal, the scope will not pick the signal up.
50mV should be easily readable, even 50uV if the signal is clean.

As for dosbox volume, it is obviously a bug.
I shall report it.

I tried several grounds, I also tried a direct connection and adding a custom load in parallel with the probe. Like I said, the signal is readable but it doesn't reach the threshold where it can be used by the oscilloscope's DSP. I'll use a small amount of external amplification next. Also I found a firmware update for my oscilloscope, let's see if that improves things.

Can somebody record Stunts Intro music played on YMF289 (not on YMF7xx). YMF289 seems has some resampling artifacts.

I don't have a card with discrete ymf289 on hand (only ymf719 and 7x4) but I do have one on the way from an ebay auction http://i.ebayimg.com/images/g/NRYAAOSwZ1BXcWHZ/s-l1600.jpg . I don't know if it will get here before I go on vacation in 2 days. If it doesn't then I'll be able to do it some time in the second half of July when I get back.

stamasd wrote:

Like I said, the signal is readable but it doesn't reach the threshold where it can be used by the oscilloscope's DSP.

50mV is PLENTY for a scope to cleanly read and represent the wave.
Something is not right.
What's the noise floor (V) with the probe disconnected?
Does your scope read from a signal generator a clean signal?
What about the test point (hook) on the scope, it should read clean 0.5V 1K square wave.
https://cdn.sparkfun.com/assets/1/7/7/2/f/52f … f5f028b456a.jpg

Tertz wrote:

To make such assertions (even subjectively) about pedia's examples you need preliminary to do their ABX comparision. I'm sure you did not. For my mind the difference I've heard is enough. Maybe later, when I'll get more basis to be paranoid about my mind, I'll play in ABX game with them.

This is pure ignorance and I shall ignore you in the name of science.

Tertz wrote:

Lower pitch of 289 in the comparision sample on that article. What makes the sound noticably worse. […]
Show full quote

James-F wrote:

Hear what?

Lower pitch of 289 in the comparision sample on that article. What makes the sound noticably worse.

The sample song in wikipedia is absolutely the worst you can compare by because it has down-bending notes

If it has music examples of real output made on those chips, then it's meaningful.

all created purely in ones mind

To make such assertions (even subjectively) about pedia's examples you need preliminary to do their ABX comparision. I'm sure you did not. For my mind the difference I've heard is enough. Maybe later, when I'll get more basis to be paranoid about my mind, I'll play in ABX game with them.

Ace wrote:

I notice the drop in sound pitch only with specific instruments.

But it is, anyway. You hear it. I hear it. While James-F theorizes it's impossible. 😀 Even after he've heard the difference on pedia's example, he still thinks it's just the delusion of his mind.

It's hard to take one sample seriously, especially one we know nothing about the circumstances of. We don't know what cards the recordings come from. It could be from two different people with different recording equipment, never mind the settings; less input gain on the YMF289's part would make it strain more, that's what the difference sounds most like to me. I don't hear the same when comparing Ace's samples - any difference I thought I heard wasn't real enough to show in my ABX'ing.

James-F wrote:

50mV is PLENTY for a scope to cleanly read and represent the wave. Something is not right. What's the noise floor (V) with the p […]
Show full quote

stamasd wrote:

Like I said, the signal is readable but it doesn't reach the threshold where it can be used by the oscilloscope's DSP.

50mV is PLENTY for a scope to cleanly read and represent the wave.
Something is not right.
What's the noise floor (V) with the probe disconnected?
Does your scope read from a signal generator a clean signal?
What about the test point (hook) on the scope, it should read clean 0.5V 1K square wave.

The oscilloscope works fine (I know how to test and calibrate them, been using them for 20 years) but its DSP isn't very good. It needs a certain signal strength to kick into FFT mode. Below that threshold it will display the signal but not do any math on it. In fact I'm not even sure it has a real hardware DSP on-board, it may all be in the software that I use with it on the laptop. I haven't found a way to force it to do complex analysis on low-strength signals.

firage wrote:

I don't hear the same when comparing Ace's samples - any difference I thought I heard wasn't real enough to show in my ABX'ing.

The majority of people don't realize how important an ABX test is and prefer to stay with their opinion out of self righteous reasons (mostly audiophiles) or because they spent more money and afraid to discover the truth. 😀
It's a standard scientific protocol when it comes to audio testing and I really appreciate a person who knows and respects this.

Please watch this and have a laugh: 😁
https://www.youtube.com/watch?v=YFKT4jvN4OE

ABX or STFU... 🤣

firage wrote:

It could be from two different people with different recording equipment

The probability of inappropriately recorded samples is low. Those people should to understood what they did not lesser than we here.
Also, I'd want to compare comlex OPL3 music in Dune (palace, intro, etc). Better on normal 289 card, not later YMF.

I don't hear the same when comparing Ace's samples - any difference I thought I heard wasn't real enough to show in my ABX'ing.

Ace said the noticable difference may depend on instruments used. Not all music may to have it.

He said, she said, I thought, you thought....
You must understand that biasing is subconscious and unwillful with the slightest suggestion or personal preferance however small it may be.
It's ABX or nothing when it comes to audio comparison.

stamasd wrote:

...and another problem seems to be that the Adlib Tracker II which I was planning to use for the tests doesn't seem to actually work correctly with a real OPL3 chip! 😒

I does work with my Yamaha cards (YMF719, 724, 744) and MELT.AMD sounds great; it works correctly with a CS4237 card, an ESS1868, and even with the CQM on a SB32.

Yup, just did some more testing again with Adlib Tracker II and the 2 cards with real OPL3. The same happens with both, instead of music I get the sound of a squeaky door essentially.

And the problem is the tracker problem itself not the cards. They work as mentioned above in games with AdLib music.

So I did the following: in DosBox I played MELT.AMD using AdLib Tracker II and recorded the OPL commands it sends to a .dro file. Then I used Droplay.exe Patch to improve OPL captures to play back the .dro file.

In Dosbox it sounds exactly like the original.

In real DOS with a sound card with real OPL3, the .dro file sounds like it should. Playing the song through the tracker again gets me the squeaky door sound. Then playing again the .dro through Droplay gets me music.

I'll see if I can contact the autor of AdLib Tracker II to let him know of this problem.

James-F wrote:

It's ABX or nothing when it comes to audio comparison.

Well in that case, I've decided to go ahead and test it out myself using the ABX test plugin in Foobar2000. This is only one test with one clip (X-Wing CD SETMUSE) with the SoundBlaster 16 CT2290 and YMF719 (I'm not using the Vibra16 CT2260 for this as I would immediately recognize the card's noisy output), but the results I got from the ABX tester are as follows (I'm using 8 runs):

1foo_abx 2.0.2 report
2foobar2000 v1.3.10
32016-07-12 14:37:47
4
5File A: YMF719 - X-Wing CD SETMUSE test.flac
6SHA1: aeffbbbd6fae6541745e9848be1e52a67ead8366
7File B: SoundBlaster 16 CT2290 (CT1747) - X-Wing CD SETMUSE test.flac
8SHA1: 298b1a38952d0e0ff1570e0cdc8101800ec74854
9
10Output:
11DS : Primary Sound Driver
12Crossfading: NO
13
1414:37:47 : Test started.
1514:38:55 : 01/01
1614:39:24 : 02/02
1714:39:46 : 03/03
1814:40:18 : 04/04
1914:41:02 : 05/05
2014:41:21 : 06/06
2114:41:55 : 07/07
2214:42:27 : 08/08
2314:42:27 : Test finished.
24
25 ---------- 
26Total: 8/8
27Probability that you were guessing: 0.4%
28
29 -- signature -- 
3005e97d66897b0263d864ba258549271ff11f441b

If you don't believe this is real, you can check the signature here: http://www.foobar2000.org/abx/signaturecheck

At least with this clip, I am able to tell the difference. With clips like Doom, however, I'm sure I wouldn't be able to tell the difference as I couldn't even tell the difference going back and forth in Audacity.

There will be more tests later on with my other clips.

Yeah, I spotted a sort of amplitude difference in the vibrating bass fret noise sound on the first beat. The buzz there is a little more forward in the YMF719 sample and I can definitely tell them apart by that.

1foo_abx 1.3.4 report
2foobar2000 v1.2.9
32016/07/13 02:06:56
4
5File A: C:\Users\Joni\Desktop\SoundBlaster 16 CT2290 (CT1747) - X-Wing CD SETMUSE test.flac
6File B: C:\Users\Joni\Desktop\YMF719 - X-Wing CD SETMUSE test.flac
7
802:06:56 : Test started.
902:07:18 : 01/01  50.0%
1002:07:39 : 02/02  25.0%
1102:07:45 : 03/03  12.5%
1202:07:52 : 04/04  6.3%
1302:08:03 : 05/05  3.1%
1402:08:20 : 06/06  1.6%
1502:08:45 : 07/07  0.8%
1602:08:56 : 08/08  0.4%
1702:08:58 : Test finished.
18
19 ---------- 
20Total: 8/8 (0.4%)

Ace wrote:

This also includes some OPL2 samples from the YM3812 on a 1990 revision of the AdLib. Also, the Vibra16 I used is VERY noisy (I was going to use a SoundBlaster Pro 2.0 CT1600, but decided not to because of its extremely heavy low-pass filter that the SoundBlaster 16s, YMF719 and AdLib don't have, which makes the sound much more muffled - it's my sound card of choice for DOS, but for consistency's sake, I left it out of this set of samples).

I believe you can disable the SB Pro's low-pass filter in the MIXERSET program.

You can, but if I remember correctly this applies to the digital part only?

James-F wrote:

The chances an average person will hear any difference in pitch while sequentially listening to a complex music track with 1Hz pitch shift is slim to none.

* Anything that is not a double blind listening test is meaningless in the audio comparison world.
Claims like "I definitely can hear a 1hz pitch difference between two full compositions played in sequential order", is 100% biased and false.

i am no musician but totally with you on this. 🤣
the only major drawback that i see the ymf71x cards is their build quality, i have never seen one looking remotely as well built as the sw1000xg or the diamond mx200.