07930: shinobi2: Missing Instrument and bad helicopter sound.

ID	Category [?]	Severity [?]	Reproducibility	Date Submitted	Last Update
07930	Sound	Minor	Always	Apr 4, 2021, 19:30	Apr 21, 2021, 22:39

Tester	Emuissues	View Status	Public	Platform	MAME (Official Binary)
Assigned To	aaron	Resolution	Fixed	OS	Windows 10 (64-bit)
Status [?]	Resolved			Driver	sega/segas16b.cpp
Version	0.230	Fixed in Version	0.231	Build	64-bit
		Fixed in Git Commit	782d827	Github Pull Request #

Summary	07930: shinobi2: Missing Instrument and bad helicopter sound.
Description	The sound driver of Shinobi for System 16B (shinobi2 rom) is not accurate. The cymbals sound of the drum set are not playing correctly and the helicopter sound on Mission 2 is "corrupted".
Steps To Reproduce
Additional Information	You can see this video for a comparison:
Github Commit

Flags
Regression Version
Affected Sets / Systems	shinobi2

Attached Files	shinobi2.inp (117,443 bytes) Apr 5, 2021, 10:10 Uploaded by Emuissues

No.18650 aaron Developer Apr 5, 2021, 08:32	Thank you for the report. Are you able to test with the latest from github? I have made a few fixes that might help at least the helicopter sound. Or do you have an INP that plays through to get to the helicopter part?
No.18651 Tafoid Administrator Apr 5, 2021, 09:57	The helicopter sound can also be accessed by using the Service Mode Sound Test .. Entry 1E To me, that does sound corrected in current Git, but I'm far from an audio expert. It surely beats 0.230's presentation.
No.18652 Emuissues Tester Apr 5, 2021, 10:13	Yes, in the current Git it's fixed. Very thank you. I've uploaded the .inp btw. Is there any possibility to fix cymabls sound to match the pcb?
No.18656 aaron Developer Apr 6, 2021, 00:03	Thanks for confirming the helicopter. Haze mentioned the cymbals could be due to the netlist filtering. Are you running the System 16A shinobi set or one of the System 16B versions? 16A versions = shinobi, shinobi1, shinobls, shinoblb 16B versions = shinobi2, shinobi2d, shinobi3, shinobi4, shinobi5, shinobi6 I wonder how the sound differs between the two versions. Only the 16B has netlist filtering, so if you run one of the 16A sets it won't have the filtering.
No.18657 vidpro1 Developer Apr 6, 2021, 12:33	I can confirm on MAME 230 Waiting helicopter sounded funny after all hostages freed. I'll try the fixed one.
No.18658 Emuissues Tester Apr 6, 2021, 12:41	I'm running System 16B, shinobi2 set.
No.18661 Augusto Tester Apr 6, 2021, 20:18 edited on: Apr 6, 2021, 20:20	how for test disable netlist filtering ? I have noticed even in Windows and Linux that using portudio if using CPU Power Profile PowerSave even having enough CPU time for emulation happen the problem that is being said above with pops in sound for some games mainly "old" games using AY and YM sound chips and strangely when accessing OSD menu plugins happen very much pops in sound and if accessing others menus not problems in sound. Several version before the option "resample if sample rate not match" was removed from mame user configuration and not removed from mame sound core, but internally enabled for good. Perhaps anything related with "resamplefilter" ?
No.18672 Emuissues Tester Apr 11, 2021, 23:37	The cymbals issue happen in the 16A version too... I doubt it's caused by the filtering.
No.18681 abelardator2 Tester Apr 12, 2021, 17:32	I add recording of music and sound effects from PCB capture. MUSIC: https://drive.google.com/file/d/1qfCHLjNTmJC2wGyLnq4j_Zp-sCF5Bexp/view?usp=sharing SOUND EFFECTS: https://drive.google.com/file/d/170VFaRo2m_NccMZrnwoMp3nyGUomG5YN/view?usp=sharing
No.18682 Emuissues Tester Apr 12, 2021, 19:45	You can notice the issue from minute 1:03 onwards of the video that I posted.
No.18683 aaron Developer Apr 12, 2021, 23:41	Can I access the music track from test mode? If so, which one is it?
No.18684 aaron Developer Apr 12, 2021, 23:42	abelardator2, thanks for the recordings. Are those during gameplay or from test mode?
No.18685 Emuissues Tester Apr 13, 2021, 00:17 edited on: Apr 13, 2021, 00:17	aaron, you can access to the music track list pressing F2, and when enter in service mode, press F2 again. From now on you can hit the attack button CTRL in mame by default to change the music tracks. The track nº5 is a good example of cymbals sound. Thank you for looking into this.
No.18687 abelardator2 Tester Apr 13, 2021, 09:29	Aaron, yes, I accessed the test mode in the PCB and played the sound consecutively.
No.18714 StHiryu Tester Apr 18, 2021, 15:30	Before delving further into this problem, I would test if cymbals are present in all emulated Shinobi sets (6 differents). It is not the first time that supposed bugs are actually features present in certain sets and absent in others. For example, the climatic effects in the Turbo Out Run appear differently depending on the version of the board (07589).
No.18715 Emuissues Tester Apr 18, 2021, 17:00	All sets have this problem. I tested myself before doing the video.
No.18727 aaron Developer Apr 21, 2021, 09:35 edited on: Apr 21, 2021, 09:37	Ok, I think I have this fixed now. The envelope was programmed for instant attack, super fast decay, but with a sustain level of 0. We were allowing the decay to happen for one tick in this situation when it should have stuck at minimum attenuation. Now the cymbal sound has a bit of pitch, which it was missing before. Let me know if this sounds closer.
No.18730 Emuissues Tester Apr 21, 2021, 20:42	Aaron, you are breathtaking, sir. Yes, it is fixed now. I don't know if its 100% accurate but at least for me it totally is. Now Shinobi in mame presents a better emulation than sega's astro city mini thanks to you. Thank you endlessly.
No.18731 aaron Developer Apr 21, 2021, 20:56	Thanks for confirming. Here is a comparison of the isolated sound: https://aarongiles.com/for/mame/shinobi-compare.wav
No.18733 Emuissues Tester Apr 21, 2021, 21:10	Outstanding difference!! Thank you. Really.
No.18734 enik2 Tester Apr 21, 2021, 21:37	Aaron, when you have some time could you take a look at the aliasing sound problem of the SMS port? https://mametesters.org/view.php?id=5830 Not FM related, but the problem persisted after the 0.225 sound sample rate conversion and mixing changes.