New Documentation: An authoritative reference on the YM2612

[notaz]

Tried to add handling for this into MAME core. It took a while to figure out that the behavior Nemesis described is not what actually causes the 'blip' sound, the blip comes from operator1, and the described operator2 behavior eliminates the 'blip'

For some reason I'm getting the best results when I use fnum=7ff and blk=5 settings as a maximum. I've checked as many games as I could:

Code: Select all

+- Ariel - The Little Mermaid (intro, 1st level)
+- Battletech (music 03, 07, 12)
+- Berenstain Bears', The - Camping Adventure (music 1, 4, ...)
+- Comix Zone (tracks 5, 10)
+- Flashback (bgm3)
+- Shaq Fu (most of the music)
+- Spider-Man - The Animated Series (intro music)
+- Vectorman 2 ("fall dig" track)

They sound fine, but they all seem to use the same settings (blk=4, mul=12, ...), probably because they all use GEMS.

[Nemesis]

I've confirmed the presence of the CSM mode in the YM2612, and whatsmore, Kega doesn't emulate it. Hooray, I've found something new!

I've still got more tests to run before I can post a full description of how the CSM mode works, and I'm not going to have time to complete them all tonight, but I'll post the details as soon as I've confirmed the behaviour.

btw, is this a bug with ch3 only or did you reproduce it on other channels ?

I've now confirmed that the detune overflow bug occurs on all channels.

[Eke]

cool, did you found any games using this ?

Btw, you already made a great discovery by explaining the Detune bug, I still can not understand how this could have remained like a dirty secret so long

sharing informations is the way to move on, especially when some people don't have a way or enough knowledge to make test on real hardware themselves... thanks a lot Nemesis

[Nemesis]

I haven't looked for games that use this feature, I've just been experimenting with my own code. If anybody wants to hunt for games which might make use of it, CSM mode is active when the channel 3 mode bits are set to "10", or in other words, when register $27 is set to 10******. CSM mode is not active when the upper bits are set to 11.

[notaz]

Looking forward to "Undocumented YM2612 features" doc

[Eke]

Yep, that should do it in theory.
As you said the YM2612 is probably using internally a non signed fixed size register and in some case it underflows. Your fix should emulate it correctly =)

well, it doesn't
I'm probably missing something with this phase increment stuff (this is definitely one of the thing I'm still not sure to understand in the emulator code)

notaz: what did you do exactly to fix the MAME core ?

[TmEE co.(TM)]

What is this CSM mode ?

[Nemesis]

Yep, that should do it in theory.
As you said the YM2612 is probably using internally a non signed fixed size register and in some case it underflows. Your fix should emulate it correctly =)

well, it doesn't
I'm probably missing something with this phase increment stuff (this is definitely one of the thing I'm still not sure to understand in the emulator code)

I added the following lines to CALC_FINC_SL in Gens to approximate the effect when I was doing testing:

if(SL->Finc < 0)
{
finc = FINC_TAB[0x7FF] >> (7 - 0x7);
SL->Finc = (finc + SL->DT[kc]) * SL->MUL;
}

Based on what notaz said, you might find the result sounds closer if you set block to 5 instead of 7.

What is this CSM mode ?

CSM mode is another special mode for channel 3, where Timer A automatically does key on/off for the channel.

[TmEE co.(TM)]

I can't make use of CSM mode in my sound engine then... Timer A is used for SFX, Timer B for BGM, VBLs are not touched.

[Chilly Willy]

What is this CSM mode ?

It's a mode meant for formant speech synthesis. One of the better methods of synthesizing speech is to break it into the addition of resonant frequencies that last a specific time. These are called formants and if you add the correct ones, you get a phoneme. You might want to check the proper wiki pages for more info.

http://en.wikipedia.org/wiki/Speech_syn ... _synthesis
http://en.wikipedia.org/wiki/Formant

[TmEE co.(TM)]

For speech synthesis I use samples... 50K of samples are enough for well understandable text only limitation of my synth : you need to be Estonian to make it talk (all languages that have "say what you read" and no pronounciation stuff will do. in Estonian "a" is spelled exactly the same in any word you encounter not like in English where the pronounciation depends on word).

[TulioAdriano]

It would work well for Portuguese, though, and most romance languages, except for French which has lots of irregularities... interesting, very nice.

Btw, I haven't seen any concrete report on the famous channel 3 special mode which would allow individual operators to be set on different frequencies.

As far as I know most tests on that feature failed to succeed, maybe wrong documentation again?

[notaz]

what did you do exactly to fix the MAME core ?

I changed the start of refresh_fc_eg_slot() to this:

Code: Select all

INLINE void refresh_fc_eg_slot(FM_SLOT *SLOT, int fc, int kc)
{
        int ksr, fdt;

        /* (frequency) phase increment counter */
        fdt = fc+SLOT->DT[kc];
        /* detect overflow */
        if (fdt < 0) fdt += fn_table[0x7ff*2] >> 2;
        SLOT->Incr = fdt*SLOT->mul >> 1;

...

Note that this only works when sampling @ 44.1hHz, so it's probably wrong. Multiplier may have something to do with all this too.

[Shiru]

Btw, I haven't seen any concrete report on the famous channel 3 special mode which would allow individual operators to be set on different frequencies.

As far as I know most tests on that feature failed to succeed, maybe wrong documentation again?

What you mean 'failed'? This mode works well both in emulators and on real hardware, some games use it. I even added basic support for this mode in TFM MM long ago.

[Mask of Destiny]

Btw, I haven't seen any concrete report on the famous channel 3 special mode which would allow individual operators to be set on different frequencies.

As far as I know most tests on that feature failed to succeed, maybe wrong documentation again?

What you mean 'failed'? This mode works well both in emulators and on real hardware, some games use it. I even added basic support for this mode in TFM MM long ago.

My guess is he's thinking of the tests as to whether or not Channel 3 special mode also worked on Channel 6.