libretro/scummvm
1
0
Fork 0
mirror of https://github.com/libretro/scummvm.git synced 2025-02-08 03:47:32 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

SCI: (CMS driver) - minor sound improvement

Browse Source 
This reintroduces two original bugs that I found too nasty to implement when I made the driver. These are present in all SCI0 CMS drivers. They do actually influence the sound in an positive way and make it more faithful when comparing to DOSBox. I have added an #ifdef option so it can be easily enabled and disabled for testing.
This commit is contained in:
athrxx 2020-05-23 17:19:34 +02:00
parent f2ec6cbf63
commit 77dc9dfd09
1 changed files with 24 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

25
engines/sci/sound/drivers/cms.cpp
View File

@ -31,6 +31,8 @@
#include "sci/resource.h"
#include "sci/util.h"
#define SCI0_CMS_ORIGINAL_BUG 1
namespace Sci {
class MidiDriver_CMS;
@ -115,6 +117,9 @@ private:
uint8 _envSLI;
uint8 _envPAC;
uint8 _envPA;
#ifdef SCI0_CMS_ORIGINAL_BUG
static uint8 _envAR1;
#endif
uint8 _envNote;
uint8 _envSSL;
@ -294,9 +299,16 @@ const int CMSVoice::_frequencyTable[48] = {
242, 246, 250, 253
};
#ifdef SCI0_CMS_ORIGINAL_BUG
uint8 CMSVoice_V0::_envAR1 = 0;
#endif
CMSVoice_V0::CMSVoice_V0(uint8 id, MidiDriver_CMS* driver, CMSEmulator *cms, SciSpan<const uint8>& patchData) : CMSVoice(id, driver, cms, patchData), _envState(kReady), _currentLevel(0), _strMask(0),
_envAR(0), _envTL(0), _envDR(0), _envSL(0), _envRR(0), _envSLI(0), _vbrOn(false), _vbrSteps(0), _vbrState(0), _vbrMod(0), _vbrCur(0), _isSecondary(id > 7),
_vbrPhase(0), _transOct(0), _transFreq(0), _envPAC(0), _envPA(0), _panMask(_id & 1 ? 0xF0 : 0x0F), _envSSL(0), _envNote(0xFF), _updateCMS(false) {
#ifdef SCI0_CMS_ORIGINAL_BUG
_envAR1 = 0;
#endif
}
void CMSVoice_V0::noteOn(int note, int) {
@ -399,8 +411,15 @@ void CMSVoice_V0::update() {
--_envPAC;
break;
} else {
#ifdef SCI0_CMS_ORIGINAL_BUG
// This is bugged in two ways. The attack rate value is misinterpreted as a int8 for the comparison (the only place in the code
// where it does that). And it always uses the attack rate of voice no. 1 as a modifier instead of the voice's own attack rate.
// Keeping these bugs will result in faithful audio. It probably even sounds as intended, since the sequencer will have likely
// suffered from the same bug and calculated the deltas based on that...
_currentLevel = ((_currentLevel >> 1) > (int8)_envAR)?((_currentLevel >> 1) - _envAR1) & 0xFF : (_envAR - _envAR1) & 0xFF;
#else
_currentLevel = ((_currentLevel >> 1) > _envAR) ? ((_currentLevel >> 1) - _envAR) : 0;
//_currentLevel = ((_currentLevel >> 1) > (int8)_envAR) ? ((_currentLevel >> 1) - _envAR1) & 0xFF : (_envAR - _envAR1) & 0xFF;
#endif
_envState = kAttack;
}
// fall through
@ -539,6 +558,10 @@ void CMSVoice_V0::selectEnvelope(int id) {
_vbrCur = _vbrMod;
_vbrState = _vbrSteps & 0x0F;
_vbrPhase = 0;
#ifdef SCI0_CMS_ORIGINAL_BUG
if (_id == 1)
_envAR1 = _envAR;
#endif
}
const uint8 CMSVoice_V0::_volumeTable[176] = {