It is possible implementation of DSOAl in Linux (alsa work as directsound) for HW acceleration of EAX at Creative sound blaster cards ?

[atirage21]

It is possible implementation of DSOAL in Linux (alsa wotk as directsound) for HW acceleration of EAX at Creative sound blaster cards ?

Or what is missing from creative company for realising linux version of DSOAL ?

It would be also nice cowork with valve due to proton emulator windows games in linux through steam client: https://www.reddit.com/r/SteamPlay/comments/k0rs46/using_dsoal_with_steam_proton/

[mirh]

Previously in this series:
#99 and kcat/openal-soft#692

And yes this could very much be integrated into wine itself directly (there has been patches in staging for a decade) but upstream is actually giving very few damns about it and in fact regressed the thing.

[kcat]

DSOAL is independently developed and Creative isn't involved in any way, there's nothing from them holding back a Linux version. It won't provide hardware acceleration on Linux since there aren't any hardware accelerated OpenAL Linux drivers. The latest DSOAL builds also require the ALC_EXT_direct_context extension which only OpenAL Soft supports (as far as I know), so hardware drivers probably wouldn't work even if they existed.

Generally speaking, a Linux version could probably be made by replacing the parts that interact with the Windows API, but there are issues where it wouldn't be a drop-in replacement; it's not an insurmountable problem, but DirectSound relies on GUIDs and HWNDs from the Windows system, and there would also be potential compatibility issues regarding parameters and structures using long types since they aren't always 32 bits like on Windows (should they stay as long types and change size, or should they change types and stay the same size? different apps would likely prefer different answers). Ultimately, I can't imagine it'd be too useful since Linux apps don't use DirectSound, and new apps aren't still using that API. If you're able to change the source for an old DSound app, it would probably be worth switching to OpenAL along with removing/replacing all the Windows calls.

It would be most useful in Wine/Proton with existing Windows apps, where the Windows build already works (barring any bugs).

[mirh]

I assume that's what they meant.

Btw is ALC_EXT_direct_context really a big deal code-wise? Like, would it be too much of a burden/duplication if it was optional?
Of course you wouldn't get its benefits, but after having ALchemy work with openal-soft, it would be funny for ct_oal to run dsoal.

[kcat]

Btw is ALC_EXT_direct_context really a big deal code-wise? Like, would it be too much of a burden/duplication if it was optional?

There could be some non-negligible performance concerns without it, since it would need to check and update the current context when making OpenAL calls, along with compatibility problems for apps that mix DSound and OpenAL since the current context is global (the direct context extension allows specifying the context directly with OpenAL calls, avoiding the need to continually check/change the current context).

There may be a couple other issues with other drivers, namely being able to open specific devices by specifying a stringified GUID (may or may not work in various implementations; the spec allows implementations to accept whatever strings it wants alongside the enumerated names, but there's no guarantee it'll support anything beyond the enumerated names), and AL_EXT_STATIC_BUFFER, which some other implementations do support though I don't know if hardware does. These wouldn't be impossible to work around, but would take some effort and introduce potential compatibility problems (the device names and GUIDs not matching what other system APIs report, or a high read-ahead for streaming buffers).

[atirage21]

From wine 9.4 to 10.0 was solved many things for renderering 3D position through dsound.

https://gitlab.winehq.org/wine/wine/-/releases?before=eyJyZWxlYXNlZF9hdCI6IjIwMjQtMTEtMjIgMjI6MTM6MTYuNDQ4NzE5MDAwICswMDAwIiwiaWQiOiI4NzQifQ

[mirh]

Eh? Not really?
Last time anything big happened to dsound was in 8.9. Then there was a small improvement in 10.0 and 9.0 but nothing too crazy.
Still no revert of the "Vista-ification" commit in 6.22.

[E-A-X]

After spending like several days by now i guess the entire EAX Support and native hardware audio processing will never make a return at all i guess?

Ive spent multiple days by now working around with OpenAL, DSOAL etc.. trying to make the old games run the EAX reverb Effects but i dont get it done.
At the very best i get audiocrackles and/or massive audiolatency and thats it.
I have already installed Linux because i wanted to play old video games aka Retrogaming as Windows 11 is no longer capable due to extreme audiolag to even playback any sound properly at all.
Even in latest W11 Games theres noticeable audiolag and horrendously useless 3D Audio Effects in like 99% of all games.

https://www.youtube.com/watch?v=J_tLvXdLKHU

Testet it here with Linux/Bazzite. I dont get it to run accordingly at all.

[mirh]

As I'm sure I said in the linked threads, hardware audio processing is a total red herring.
It's not technically impossible (although I don't even remember if ctxfi/emu10k1 are basic feature complete and stable to begin with), but it's 2025 and it's unneeded.
On top of the fact that for the best quality that openal-soft can pull off, they probably couldn't even support it all. At best its selling point could be helping with RE and whatnot, but eh.. at that point you could just straight up help here like @bibendovsky did.

Secondly, it's kinda odd that for the problems in linux you seem to be blaming yourself, while for the ones in W11 you already think the culprit must be the OS? Anyway, in both cases the problem is likely that you haven't override the CLSIDs.

And thirdly I don't see the connection between the first point and the second point.

[E-A-X]

Because under W11 the audiolatency is so high, that it doesnt function in any old game anymore out of the box.
Adding EAX will add 0.5-1.5s of audiodelay. Every source is either far away from its origin, not audible or lags super hard when changing camera fast.

That said, Playstation 5 does it, they use Tempest 3D Audio and in well made products, such as Gran Turismo 7, it just melts the competition in terms of audio.
The realtime calculations are excellent and audio is never boring to listen.
Meanwhile software products even with the use of Wwise dont sound good at all.
Worst is the lack of audiosources simultaniously.

Dedicated hardware audio chips are still required to this day as normal software codec running on a CPU thats absolutely inefficient in its design to render audio is not good.
Every game results in a playback of maybe 3-4 soundsources in total, meanwhile games used to playback up to 512 sources at once loud and clear.

My CPU has pretty much no processing power left after using 100% of it for 480Hz screen.
And what the CPU does is highly inefficient by architecture.
These original 400MHz soundchips did, what required a entire playstation 4 pro CPU just for audio processing.
With that, theres no more headroom for rendering frames.

As of right now, the only option available to return EAX and proper gamesound that plays back as it was designed is to run a Retro XP machine with a XFI Soundcard.
It does no longer function correctly since Vista and it has reached its peak lows with Win11 by now.

I used to play these games in and out back in the day with a XFI card and ever since im running my nickname.
If i check up any Creative Alchemy/OpenAL workarounds not running natively on XP, i hear the differences of the processing as they are inaccurate and worse.

What made EAX so unique was that you could take any headphone and any budget logitech 2.1 system and the reverb would 100% accurately reproduce the directional audio like a 7.1 system with ease, on top of that creating reverb of the gameworld, not useless room acustics.
Closing your eyes and get that crosshair placed directly on the source with 100% accuracy.

I havent seen any tech remotely competing with that ever since. Especially not for a price for 100€.

Even if running games via Linux, footstep audio has less audible range than native experience and are much more difficult to locate.
HRTF is like 30% of EAX, dolby headphones, DTS X, CMSS3D, thats surface scratching, lack of realtime reverb processing of any room or area and not having that echo just removes massive amount of audible information.

Those are original native experiences:
https://www.youtube.com/watch?v=jptoFrSnvdY

https://www.youtube.com/watch?v=819pyCVMtCQ

Can close your eyes, you dont need a screen anymore. Basicly Audio was that precise back in the day, it expanded your 4:3/5:4 CRT Screen to a 360° virtual screen. There wasnt a single deadspot in the acoustics.
If you would take that Bioshock Remaster, it would be 2% of the original experience, a complete waste of money due to the lack of EAX, HAL, D3D etc...

[kcat]

After spending like several days by now i guess the entire EAX Support and native hardware audio processing will never make a return at all i guess?

OpenAL Soft and DSOAL support the EAX API, and has been successfully used in native Windows as well as Wine/Proton.

Hardware audio processing is a little more nuanced. It's very unlikely there will ever be dedicated hardware audio cards again, like the Creative cards of the past. In specialized devices, you may get a separate processor that's intended to focus on audio work, though at least in game consoles, I don't think they've had that since the PS1 era (maybe the XBox360 had something?). It's more likely that audio work is done on a processor or cores that are shared with other work. In modern times it is possible to offload some work from the main CPU to the GPU, such as doing raytracing to calculate reverberation and obstruction/occlusion/exclusion properties in an API-agnostic way, that can then be applied to the audio API you have which can do more work on the GPU and/or CPU. It's not dedicated audio hardware, but is that "hardware audio"? At the bottom of it all is always going to be some hardware running some code, so I guess that's a more philosophical question than a practical one.

Every game results in a playback of maybe 3-4 soundsources in total, meanwhile games used to playback 512 sources at once loud and clear.

OpenAL Soft has no issue handling a lot of sound sources. While I'm not sure what I have that can stress test it that much, it can handle well over 100 constantly updating sources no sweat, with a pseudo-dynamic reverb, a stereo track being run through a Super Stereo decoder, and a UHJ encoder thrown on the end for good measure (I can't seem to make a decent video to show it that's under 10MB to attach here, so I guess you'll have to take my word for it).

There certainly is room for OpenAL Soft to improve, but a lot of audio quality is dependent on audio design and the audio engine itself. I don't think I've seen anything that takes full advantage of OpenAL Soft's capabilities (fully dynamic multi-zone reverb, source obstruction/occlusion/exclusion, volumetric ambisonic sound sources, ...), and there aren't many games that utilize EAX4 hardware features, let alone take any EAX4 hardware to its limits.

[E-A-X]

Playstation 3 has had a dedicated Chip and Playstation 5. As said, not all Games use proper Audio Engines, as of why i put the GT7 example, as thats one of those rare games, that care about audio and in a racing game thats important, because if you take a different line, you want to hear different audio, if its the same over and over again, like in iRacing, Assetto Corsa, Dirt Rally 2.0 etc.. it would get boring and also annoying at some point.
PS2 as of example has none, playing Splinter Cell Pandora Tomorrow with that would immediately result in useless Audio, as it gives you zero awareness at all. Its literally meant to be played on Windows XP or Linux, everything else is just worse.

What i have figured is, that while im running on Bazzite, every single Game has way better Soundquality, than on Windows 11, right out of the box.
Like i have not an issue to hear ~50+ sounds in games like Megabonk and locate their sources freely. Meanwhile on Windows 11 the sound is full of lag, can barely hear 5 sounds total and it is completly inaccurate.
Running Valheim in example on Windows and pick up a stone out of the ground its full of lag. Audio is super far behind of the imagesource like half a second, no issue with linux, audio is in sync with the image.

The same is for games like Counter Strike 2, Audio via Windows is next to useless. It would skip spawn sounds in a deathmatch game if someone shoots his gun. Meanwhile with Linux it plays back 5-6 shooting sources, footstep audio and spawnsounds regularly, including proper directional audio of the source and its distances + its lagfree. Yet its not what it could be.

What confuses me is as of why this entire tech was thrown out of the window. Every single Windows Game is a mess out of the box on the OS. And if there are attempts of making good audio, it sounds unnatural, compressed, it causes pressure inside of the ears. Sounds that are within 5 metres of range sound like they are 200+ metres far away. Sounds at 20 metres are barely audible anymore and distances are only calculated by predefined volume changes, more distance = sound is geting lower volume - thats all.
Based on hearing it is impossible to tell the distance to an object, the size of a room, the degree of angle in height etc...

Interestingly enough, there were attempts but they also vanished.

Best comparison that exists of an "up to date" game is Dying Light 2, E3 2019 vs Retail.
https://youtu.be/38NXe4uKMrY?t=42

Retail has (in my opinion) completly awful and useless Gamesound. That sound will never generate a single feeling while playing the game, it will never shock and it will never cause the feel of intense threat moments.

Interestingly enough, when playing Arc Raiders via Proton the Audio is very very close to the original EAX Experience, however the Audioengine doesnt manage to playback multiple soundsources, as its super easy for players to sneak up, if you fire your gun. Footstep Audio of other players will vanish and not exist when theres gunfire going.

Either way, what i experience when i use DSOAL etc. is that the room acoustics do not exist. Its obvious in Max Payne 2 prerendered Cutscenes, as EAX is included in the audiofile, as soon as the cutscene ends, its vanished.

I the video above, when i do use "Native" in the Protontricks, the audiocrackles and it tries to use EAX Reverb Room Acoustics but it doesnt handle it properly. The Sounds cut out way too early and replace each other constantly.
Also theres audible lag of +1 Second from Input to Output.

I cant even figure the cause, is it because i do no longer use a X-FI Card and instead a Soundblaster Z? What did go wrong?
If i check other sources that try to get these things done, their gamesound is absolutely equal to my sound. EAX only in Cutscenes, gone when gameplay.
Most iconic are the rain effects at the hospital outside the windows, the audio is not correct. It may still sounds years ahead of current productions but not as it should be.

[mirh]

W11 may suck for a lot of reasons, but you are doing something very wrong if your latency isn't in the low double digits.
Everything else is just detached rambling (ironically enough you linked a bioshock video, which has no problems in vista).

If you want to help you should first try to fix whatever madness is going on with your situation, and then help compare games for the bugs that are surely still present in dsoal or openal (e.g. #81)

[E-A-X]

The only madness i can imagine is, that im running a 480Hz OLED Screen and with that, it makes the entire Desync from Audio to Imagesource way more obvious and that im still highly sensitive in terms of listening.
As LCD Tech is full of motionblur during quick movement, it hides the issues with ease.
It cannot be on my end, because if it was, i wouldnt be able to have 0 audiolag/image desync in one video and clearly hear desync in those videos, that run Creative Alchemy/OpenAL on newer OS.

My guess is, that the majority is entirely unsensitive to that whole topic and as growing older and treating your ears bad with loud sounds, never using noice canceling headphones etc.. at work, that they can no longer notice it.
Exactly like the majority never noticed that they purchased super awful technology aka LCD Panels and replaced their golden, motionblur free CRT Screens, that wouldnt smear out during fast inputs.

Everytime if i ask my dad if he doesnt hear his broken lightswitch and how dimming the lightbulb causes highfrequency noises, hes like "nope, i dont hear it". I would never ask him to compare gamesound aswell, as its clear, he cannot hear differences anymore due to age and beeing completly careless at work without ear protection.

So theres that too to keep in mind.