Protracker player with support for external sound fx

[phx]

Originally Posted by h0ffman

Anyways, I'll grab your latest beta and give it a try.

Thanks. But there is no beta available, just the patch. I will upload a beta of the source and tell you the URL (maybe after I had a look on the master-volume issue).

One thing I have noticed is the master volume doesn't appear to be effecting samples which are already playing.

Indeed. I never noticed that in our games, which offer a volume option in the menu, because new samples are usually playing fast enough.

I have a short fade which occurs when soundtracks are switched and the looped samples being played by the mod are not reducing until another sample is played.

Yes, that's annoying. I will have a look these days.

Is that how it currently works, i.e. master volume doesn't come into effect until a sample is triggered?

Correct. The volume is always set together with a new sample. But the last volume is stored in the channel status, so I could probably just use this and write a new volume to AUDxVOL whenver the master volume changes. Seems easy.

[h0ffman]

Legend! Gimme a shout when its ready.

[Photon]

Yes! Master volume must be changed on command not trigger because this is used in all kinds of modules to emulate arp/echoes. Sometimes also the notetempo is not fast enough to make a volume fade act fast enough (without some insane fade value requiring a - master volume reset value on the next note).

[phx]

Latest ptplayer 6.1beta is here:
http://sun.hasenbraten.de/~frank/TEST/ptplayer61beta.lha

Now all channel volumes should be changed immediately during the mt_mastervol() call.
Set NULL_IS_CLEARED=1 to test the idle-looping of samples at $0.

Complete list of changes in current beta:

- Fixed note delay command (EDx), which still played the previous note
  in some situations (Antiriad/EAB).
- mt_mastervol() must change the volumes of all channels immediately
  and shouldn't wait for the next sample being played.
- Symbol ENABLE_SAWRECT may be used to disable sawtooth and rectangle
  waveforms for vibrato and tremolo, which saves memory for their tables
  (suggested by Antiriad).
- Symbol NULL_IS_CLEARED may be used to indicate that the memory locations
  $0 and $1 in your system are zero, so they will be used for idle-looping
  of audio channels.
- Removed cia.i and custom.i include files and included the required
  symbols directly into the source.
- New function mt_loopfx() for playing looped sound effects.
- New function mt_stopfx() for immediately stopping a sound effect.

[a/b]

Not sure how much in demand it would be, but you never know...
Instead of using static tables, you could generate them with a one-time initialization. Why? Well, in case you also need the compressed player to be as short as possible.
Like in my case, I'm using a heavily modded and optimized P612 but the same thing applies, I was able to reduce its compressed size from 2.7kb to 1.7kb (trying to stuff as much as possible within a 5.5kb track, for example) just by generating the vibrato sine table and fine-tuned period table. Each routine is about 40 bytes long (and if you play "dirty" you can scrap that code so it occupies no space once it's done).
The same could be done with master volume and the other 2 vibrato tables. They don't look too complicated.

[phx]

If I would do that, then it has to be optional. ptplayer is mainly intended for games, where I would rather save some bytes in RAM than on disk.

I remember a few guys here worked on a routine to generate the tables during runtime and maybe you were one of them?
When it really works you can always send me a patch by email and I will try it.

[roondar]

Only feature request I'd have is a possibility to configure a channel that the player won't use at all (including no audio interrupts for that channel, nor any other register writes to it). This would make integrating my mixer with your new versions easier than searching through the code myself

[phx]

Originally Posted by roondar

Only feature request I'd have is a possibility to configure a channel that the player won't use at all (including no audio interrupts for that channel, nor any other register writes to it).

Doesn't mt_musicmask() work for your purpose? It makes sure that no sound effects are played on one of more channels. For the music it is up to you. Simply load a MOD which doesn't use the channel. But I have the feeling that I misunderstand you...

EDIT: Oh, I think you mean that audio interrupts are generated because such a channel is idling with a 1-word sample?

[roondar]

Hmm, maybe I'm misunderstanding. I'll check with a three channel mod to be sure and come back to you. First have to find one though (as it is, my mixer test program just plays back a 4 channel mod and I've simply hacked the player source to never touch AUD3).

[Antiriad_UK]

Originally Posted by a/b

Not sure how much in demand it would be, but you never know...
Instead of using static tables, you could generate them with a one-time initialization. Why? Well, in case you also need the compressed player to be as short as possible.
Like in my case, I'm using a heavily modded and optimized P612 but the same thing applies, I was able to reduce its compressed size from 2.7kb to 1.7kb (trying to stuff as much as possible within a 5.5kb track, for example) just by generating the vibrato sine table and fine-tuned period table. Each routine is about 40 bytes long (and if you play "dirty" you can scrap that code so it occupies no space once it's done).
The same could be done with master volume and the other 2 vibrato tables. They don't look too complicated.

Hey. Could you send me the code for the vib table if you still have it? I’ve got a minimal version of phx’s code for 64k amigaklang intros that already does the period and fine tune tables in code that we worked out in another thread but would be cool to generate more (I’ve not be able to get p61 to work with amigaklang due to the way it modifies samples, and frank’s player is nice to use anyway)

The current modified version is below, I’ll update it with phx’s beta fixes later. I’d cut the master volume as didn’t need it at the mo.

https://github.com/jonathanbennett73...er_minimal.asm

Originally Posted by phx

If I would do that, then it has to be optional. ptplayer is mainly intended for games, where I would rather save some bytes in RAM than on disk
I remember a few guys here worked on a routine to generate the tables during runtime and maybe you were one of them?
When it really works you can always send me a patch by email and I will try it.

Me . Hope you ok with me putting my modified version on GitHub with same license, mate.

[a/b]

Originally Posted by Antiriad_UK

Hey. Could you send me the code for the vib table if you still have it?

Comes with a couple of warnings...
Because of effect optimizations the table is negated (negative half of each row comes first), so you'll have to rearrange the move.b order.
This is written for speed and size, not only for size (both code and base table are placed in a disposable area and overwritten after the use, so the size doesn't matter 100%).

; a3 = FrameCtr (a.k.a. mt_Counter)
	lea	(VibTremTab\.End-64-FrameCtr,a3),a4
	moveq	#15*2,d2
.VTLoop1
	lea	(VibTremTab\.GenData-FrameCtr,a3),a2
	lea	(32,a4),a5
	clr.b	(a4)+
	clr.b	(a5)
	moveq	#64/4-1,d1		; sine with 4x mirroring
.VTLoop2
	move.w	(a2)+,d0
	mulu.w	d2,d0
	swap	d0
	move.b	d0,(32,a4)		; [pi,pi*3/2]
	move.b	d0,(32-1,a5)		; [pi*3/2,pi*2]
	neg.b	d0
	move.b	d0,(a4)+		; [0,pi/2]
	move.b	d0,-(a5)		; [pi/2,pi]
	dbf	d1,.VTLoop2
	lea	(-1-64/4-64,a4),a4	; start of next row
	subq.w	#1*2,d2
	bge.b	.VTLoop1

VibTremTab
.GenData
	DC.W	$1800,$3100,$4a00,$6100,$7800,$8d00,$a100,$b400
	DC.W	$c500,$d400,$e000,$eb00,$f400,$fa00,$fd00,$ff00
.GenDataEnd
	DCB.B	16*64-(.GenDataEnd-.GenData),0

[a/b]

OK, I found an older test version, this should probably be more useful to you since it's shorter (only 2x mirroring, but larger base table which shouldn't matter) and no negation:

	lea	(VibTremTab\.End-64,pc),a1
	moveq	#15*2,d2
.VTLoop1
	lea	(VibTremTab,pc),a0
	moveq	#64/2-1,d1
.VTLoop2
	move.w	(a0)+,d0
	mulu.w	d2,d0
	swap	d0
	move.b	d0,(32,a1)
	neg.b	d0
	move.b	d0,(a1)+
	dbf	d1,.VTLoop2
	lea	(-64/2-64,a1),a1
	subq.w	#1*2,d2
	bge.b	.VTLoop1

VibTremTab
	DC.W	$0000,$1800,$3100,$4a00,$6100,$7800,$8d00,$a100
	DC.W	$b400,$c500,$d400,$e000,$eb00,$f400,$fa00,$fd00
	DC.W	$ff00,$fd00,$fa00,$f400,$eb00,$e000,$d400,$c500
	DC.W	$b400,$a100,$8d00,$7800,$6100,$4a00,$3100,$1800
	DCB.B	15*64,0
.End

[Antiriad_UK]

Originally Posted by a/b

OK, I found an older test version, this should probably be more useful to you since it's shorter (only 2x mirroring, but larger base table which shouldn't matter) and no negation:

Thanks, I'll have a look

Edit: This is working nicely! It still seemed that the negative values came first but I just switched the order of the moves around as suggested and it matches the normal values perfectly. I'll add it to the github version shortly. Thanks

[DanScott]

Would be great to have a version of this player with all options available as equates.

I used a stripped down version for Colombia, with the SFX handling stripped out and the master volume removed too (was able to remove all those tables)

[Antiriad_UK]

Originally Posted by DanScott

Would be great to have a version of this player with all options available as equates.

I used a stripped down version for Colombia, with the SFX handling stripped out and the master volume removed too (was able to remove all those tables)

Yeah, I'll do my best to make changes configurable like that. I don't think phx will be keen on a lot of the stuff crazy 64K demo coders want, and the source is _really_ clean at the moment compared to the EQU insanity that is p61 so would be a shame to pollute the original code! So I'll just make available any changes I make as a side-project

Edit: I've updated the modified github version with phx's beta changes and a/b's sine generator. The default player (with all options/sfx turned on) is 13096 bytes. The default player with MINIMAL defined and saw/rect waves disabled is 5457 bytes. The modified player with the period/finetune/sine vibrato generation is 3116 bytes (plus 2672 bss).

[phx]

Originally Posted by Antiriad_UK

Hope you ok with me putting my modified version on GitHub with same license, mate.

Sure. If you read the license you know that it is public domain.

Originally Posted by Antiriad_UK

Yeah, I'll do my best to make changes configurable like that. I don't think phx will be keen on a lot of the stuff crazy 64K demo coders want

LOL
Yes... I would have to see it to judge how intrusive it is. There is already a lot of conditional assembly in the code (MINIMAL & Co).

The default player (with all options/sfx turned on) is 13096 bytes. The default player with MINIMAL defined and saw/rect waves disabled is 5457 bytes. The modified player with the period/finetune/sine vibrato generation is 3116 bytes (plus 2672 bss).

Wow! Nice.

[DanScott]

Just popping Antiriads version into my demo system.

I'm getting a "repeatedly defined symbol" error, as the EQU's for the hardware registers are conflicting with my "custom.inc"

Might be worth popping an "CUSTOM_REGS equ 0/1" or similar, around those (and the CIA defines)

[Antiriad_UK]

Originally Posted by DanScott

Just popping Antiriads version into my demo system.

I'm getting a "repeatedly defined symbol" error, as the EQU's for the hardware registers are conflicting with my "custom.inc"

Might be worth popping an "CUSTOM_REGS equ 0/1" or similar, around those (and the CIA defines)

Good idea.

[phx]

Originally Posted by DanScott

I'm getting a "repeatedly defined symbol" error, as the EQU's for the hardware registers are conflicting with my "custom.inc"

Why do you include your custom.inc into ptplayer.asm?

Might be worth popping an "CUSTOM_REGS equ 0/1" or similar, around those (and the CIA defines)

Another solution is to generate multiple object files and use a linker. So each module keeps its own namespace.

[DanScott]

I don't link object files I'm old-school