I have some result like that recording my sound card, and I Always fix that in audacity (it removes in 90% of the tested cases)
https://www.youtube.com/watch?v=mlKX6p85R2M

The recording sound card should use ASIO, which Audacity doesn't support. I use Sound Forge Audio Studio with an E-MU 0404 USB sound card.

If the OPL2-3 is played in Windows 98 also make sure to select speakers and best quality re-sampling in the sound card control panel. And make sure there is no clipping, although with my YMF-744 with all sliders set to max, I've not seen OPL music clip.

Additionally, mute all other inputs and outputs and place unused/muted volume sliders to the bottom.

Are these clicks present when you record analogue?

@Stretch: I'm already using ASIO but it will not change the recording quality. Changing this is just for latency. The CT3910 is on pure DOS 6.22 machine so no Windows and as i've wrote, i'm NOT using analog part, only the SPDIF output. SB16SET.EXE has NO influence on the spdif output volume. It's fully digital from gaming machine to the recording machine. And it's not clipping, the general spidf output is -30 -18 dbfs.

@philscomputerlab: yes theses "clics" are present too on analog output. Especially in LucasArts games. I got thoses nasty clics on Indy 256 VGA intro's on the CT3910, AdLib or CT1350B. By the way i don't have the "clics" at the end of LOOM intro when i'm playing the .LAA with ADPLAY. I tend to think that it's a soft/hard programming bug with certains games. I've never got "clics" when playing some modules with Adlib Tracker II for example.

Before i started to record through SPDIF, i was 90% sure 'clics" were generated by the analog part of the card, but now i know it's not. Maybe it's a logic bug in the way of programming the YM3812/YMF262. Thoses 'clics' are often present when it's only a single lead instruments played with some arpeggios. I'm pretty sure it's not a faulty card, because i got theses on at least 4 differents cards, plus it's present (not always) on Dosbox emulation too...

I'll will post some others examples later.

Hmm, these are quite old games.

Could you try disabling the L1 cache on your 486 to see if it makes a difference?

Do you have a recording of Last Crusade by any chance?

Well the clicks are on OPL output because every time a channel is keyed on the waveform phase generator is reset to zero.

And this is particularly nasty sounding on Lucas games (like Monkey Island 1) because their midi player starts a new note on the first free OPL channel, and it is possible that a note was just turned off, but it is still slowly decaying, and it is suddenly changed to play another instrument at another note. In other words, the sound does not have time to decay to silence before it is changed to another sound.

The CT3910 does not have a real Yamaha OPL chip, but it should not matter. I am just not familiar with CT OPL implementation, and not familiar with that card, so I don't know if the sound is directly digitally routed to SPDIF output or is it going through OPL DAC and then in analog form it is later on converted to digital form through ADC again. So your recording path may or may not be purely digital.

So the clicks are there anyway, regardless of any sampling rate conversions (which should be avoided anyway). Natively the OPL2 and first generation OPL3 chips have approximately 49716 Hz sampling rate, while later generation OPL3-L for example already resample to 44100 Hz and has a bit different timing anyway.

The clicks are worse on real hardware, because the sound driver writes something like over 10 OPL registers, and each write requires a delay of more than one sample, so it's actually a series of many clicks over many samples. When playing back on a emulating music player, the music player can just virtually write all the registers without delay (if it wants to) so it's possible there is only one click, a discontinuity between just two samples.

the clicks are probably due to an improper SPDIF cable. u need to use a proper 75ohm cable for proper SPDIF function or u will get these clicks
u cannot use a normal RCA cable for SPDIF u need a 75 ohm proper SPDIF Cable
even tho using a normal RCA cable will "work" it wont work *WELL*
if u dont have a proper 75 ohm SPDIF cable then u can try to use a yellow RCA video cable.. but u want a higher thicker more shielded cable. not these thin black video cables that commonly come OEM with tv's + peripherals

anyways perhaps im wrong but im pretty sure i had a very similar problem of this "click" sound when connecting a yamaha motif via SPDIF..
it happens often + randomly..for me i solved this problem by using a better quality cable

I've never heard clicks on these games, so I'm very curious for recordings...

Here's the intro of Indiana Jones & The Last Crusade (256 colors version):

WAV
FLAC

There's some markers on the wav file to locate thoses clics.
Since i got thoses clicks on analog output too, i don't think the spdif cable is faulty. It's the original spdif cable bundled with the sblive (2 pins cable for internal connector, not rca).

EDIT: i'll try to mess with ISA settings in bios, and cpu clock. Maybe the machine is too fast.

Hmm that definitely sounds off.

I don't have any recordings anymore, but it should not sound like that.

Edit:

This is the only version I found. It's on a AWE64 Gold which doesn't have an OPL3 chip.

I've listened your records. I've disabled Chorus and Reverb on the AWE32. I've compared with a real AdLib (YM3812) my record seem to be be ok (same with dosbox emulation minus the 'clicks'). I'll test with my AWE64 (non gold).

It would be interesting to see if the clicks are still there if the recording was unmodified digital, e.g. with a DigiSnap. With the current setup, the OPL's 49.716kHz output is being downsampled to 44.1kHz, then upsampled again to 48kHz, so some samples will be lost and others merged in that process.

If the clicks are there from the DigiSnap, then you know with 100% certainty that they are there on the original chip.

Yes it will be interesting. Unfortunately i don't have ths soldering skills for this and it's only work with YM3812.
Meanwhile i've changed the playing machine with a lower cpu (386sx40) and the "clicks" are less present but still it's not perfect.

The DigiSnap is easy to modify for OPL3 (well, at least YMF262) capture, just use the leftover NAND gate to invert the bit clock from YMF262. It will still be compatible with OPL2 (YM3812), because data is stable on both bit clock edges. Of course the software needs to be modified for OPL3 to store the raw capture instead of emulating the DAC analog shifter. Still, it's not stereo and needs a capturing PC old enough to have EPP parallel port, but it'll work.

Here's a screenshot of a real YMF262 (OPL3) playing neointro.hsc, a direct digital capture of serial audio bus via USB, decoded to signed PCM values and imported into audio editor. Those spikes are quite visible, as well as audible. The music is mono but as both left and right OPL3 channels play the same sound in OPL2 mode, this is not a simple one-off bit-glitch during capture or other error as it is far too deterministic, they really are the actual left and right channel samples.

I bet I have some Monkey Island captures somewhere, but I could not find them. IIRC, they have similar clicks, and Indy should have approximately similar sound driver anyway.

Thanks for your reply. After many test with different hardware, my conclusion was that the driver used by game was involved. This explain why with others games i didn't have thoses spikes or a least more less.

Jepael wrote:

a direct digital capture of serial audio bus via USB

For curiousity, how did you do that ? Custom build ?

Thanks.

DjLc wrote:

Thanks for your reply. After many test with different hardware, my conclusion was that the driver used by game was involved. This explain why with others games i didn't have thoses spikes or a least more less.

I'd blame both the chip for having the property sudden phase reset on note start, and the driver, instruments and music itself for not hiding this property.

DjLc wrote:

Jepael wrote:

a direct digital capture of serial audio bus via USB

For curiousity, how did you do that ? Custom build ?

No, just an USB logic analyzer.