Waveform's null level

[607]

I'm having a problem in MaxYMiser where when a sample starts and ends I hear a "pp". I think it happens because the samples I'm using are centered around the 0, while the Steem audio doesn't seem to be.
To see what I mean, here's a screenshot of Audacity:
Does anyone else have this problem too?

[AtariZoll]

You need to normalize it in Audacity with option "Remove any DC offset .." on.

[607]

You need to normalize it in Audacity with option "Remove any DC offset .." on.

That works and could be quite useful I suppose, thanks.
However, that isn't the problem here. I don't want to use audio from Steem in Audacity, I want to use audio from Audacity in Steem.
I'd suppose I could perhaps normalize centering on a different level, but I wouldn't know what level as it seems to change, somehow.
If you record from Steem, do you get something similar to my screenshot?

[AtariZoll]

I just did my first 2 sound recordings from Steem (usually I do it from real HW, together with video ).
Used Steem 3.2 and Steem SSE 3.808 - saved DMA audio and PSG audio. And both were miscentered, as your shot.
So, it is just smaller bug in Steem - saving with incorrect WAV flag about signing. - Actually, they should unsign it - see lower why.

Normally, WAV is with unsigned values, while STE DMA audio needs signed, 8-bit PCM data. So, you need to save, pardon export it as signed, 8-bit mono or stereo, and of course with supported sample freq. But to get proper signed signal, you must have it centered in Audacity first. And here is the problem - with 8-bit PCM no signed, just unsigned option by export as WAV. So, probably better is to save as RAW - then no WAV header too. I used earlier to save as unsigned from Wavelab, because it did not have RAW save, and signed in Atari SW self, after loading PCM data.

[607]

I just did my first 2 sound recordings from Steem (usually I do it from real HW, together with video ).
Used Steem 3.2 and Steem SSE 3.808 - saved DMA audio and PSG audio. And both were miscentered, as your shot.
So, it is just smaller bug in Steem - saving with incorrect WAV flag about signing. - Actually, they should unsign it - see lower why.

Normally, WAV is with unsigned values, while STE DMA audio needs signed, 8-bit PCM data. So, you need to save, pardon export it as signed, 8-bit mono or stereo, and of course with supported sample freq. But to get proper signed signal, you must have it centered in Audacity first. And here is the problem - with 8-bit PCM no signed, just unsigned option by export as WAV. So, probably better is to save as RAW - then no WAV header too. I used earlier to save as unsigned from Wavelab, because it did not have RAW save, and signed in Atari SW self, after loading PCM data.

I'd like to use real hardware, but MaxYMiser has problems on my STE :\
Thanks for your explanation, but I'm not sure if I know quite what to do.
At the moment I'm doing the following:
import wav, which is correctly centered. Frequency 44100Hz. Export Audio, as Other uncompressed files, with Header: "RAW (header-less)" and Encoding: "Signed 8-bit PCM".
If I now import this sample into MaxYMiser, I still get the "pp" sound when the sample starts and ends. It doesn't matter if I let the sample play completely or cut it short, I'll always hear the "pp" when it goes back to silence, as the silence level is lower than the centre of the sample.

[AtariZoll]

It may be that MaxYMiser performs signing after load. So, try to save as RAW, unsigned from Audacity.

[607]

If I export them unsigned, the "pp" isn't there. However, the sample doesn't sound like it's supposed to sound. In MaxYMiser there's a Sign/Unsign button, and if I press that, it sounds right again, but the "pp"'s reoccur.

[AtariZoll]

Well, then you need to use fade in at start and fade out at end.

[607]

Well, then you need to use fade in at start and fade out at end.

In MaxYMiser or in Audacity? I tried a fade out in MaxYMiser before, and I don't think it worked, but I'd try again.

[Steven Seagal]

Not sure I understand the real problem, but for info, samples for DMA sound are signed, so we need to adjust to their "zero" (128) or we'll get "pops" when sounds start and stop (there may be little pops anyway).
But the PSG emu is unsigned, its zero is 0.
Mixing unsigned and signed is trouble.

[AtariZoll]

The issue is not how it works in Steem when playing audio, but only the WAV saving. WAV with 8-bit signed is just not used format, and both STE DMA and YM sound is saved as signed 8-bit WAV. So, all what should correct is to unsign them when saving audio. Concrete, just add $80 to every byte in data stream - bytewise, of course.

[Steven Seagal]

But then it would sound totally wrong. It is saved as it is played, unsigned when 8bit, signed when 16bit, which is the rule for WAVE PCM (I think).

[AtariZoll]

I tried 16-bit WAV - and problem is exactly the same:
So looks saved WAV:

SteemSav.jpg

And after DC offset removal:

AfterDCoffsRemoval.jpg

This was simple PSG music. And it sounds exactly same played on PC. But upper is just wrong, and on some systems can make problems - what is actually why this thread is started.

[Steven Seagal]

PSG emu uses ljbk's table, with unsigned 16bit values.
Steem must convert to signed, by doing XOR 0x8000.
Waveform is the same, zero is changed.
I'm no expert in DSP though, I'll look into it again.

[607]

I set up Hatari today, and did the same stuff there. There were no problems. There aren't any problems like this on my actual ST either.
So it might be a Steem exclusive bug.
I didn't find it in the bugs list, so it may need to be reported.
I now know how to work around the problem, but it'd still be better if it'd would work as expected in Steem, too.

[Steven Seagal]

So it might be a Steem exclusive bug.
I didn't find it in the bugs list, so it may need to be reported.
I now know how to work around the problem, but it'd still be better if it'd would work as expected in Steem, too.

If you think there's a bug AND you use Steem SSE, please fill in a precise report there and I'll check it.