Audio Tweaking (dev branch)

[DnA 5143]

The audio changes which delayed Development branch updates are now partly published. As with most of these large tasks, this is ongoing work and so the current state is not considered final (in some places it may be worse for a little while). Effectively the core foundations are in place and now each individual asset needs tuning and fixing. We feel it will improve the soundscape in the final game :622:

Senior audio team member Ondřej Matějka goes over the details:

We’ve added a so-called limiter effect. If there are a large number of shots at the same time, there was an unpleasant effect of 'cracking'. The limiter should help counter this eventually.

We’ve changed the audibility of some sounds based on distance. Due to the changes to the sound engine it is now possible to affect the sound very accurately. As example, by separating the samples for weapon shooting and their mechanical parts, you can now not as easily hear the shooter in a remote area. With this change, the player also gets a better sense of ambient sound, because e.g. crickets are now heard only in a limited space. Another big change is the accurate setting of the volume of resources (dB) and controlling the audio mix (vehicles versus weapons for example). Rather than tweaking volume and loudness in the source samples, these are now all normalized and tweaking is done in configuration.

Finally, we’ve introduced an improved WSS format. With this change we can drastically reduce the size of the sounds by more than 60 percent and format WSS finally supports stereo samples.

Of course, there have been many fixes and tweaks to sounds themselves based on your comments. Please help our audio team by continuing to report possible improvements in this field.

[kremator 1065]

Good to see new stereo sounds. Looking forward to hearing the new soundscape!

[bigpickle 0]

Hi DNA and thanks very much for the info, I have a couple of questions about the new sound works if you think you may be able to ask the sound guys.

format WSS finally supports stereo samples

Will we get a tool to support this or does it mean the engine now processes wss in stereo format?

We’ve added a so-called limiter effect

Will this be like arma 2 were the amount of samples is capped and going over that cap stops the first of the sound files from playing.

Many regards & Thanks for the communication

Bp

[DnA 5143]

Will we get a tool to support this or does it mean the engine now processes wss in stereo format?

Currently it means the engine processes it. I'll investigate whether we can add a conversion tool to the tools package.

The rest will need to be answered by my team :)

[NeuroFunker 11]

Hi DNA and thanks very much for the info, I have a couple of questions about the new sound works if you think you may be able to ask the sound guys.

Will we get a tool to support this or does it mean the engine now processes wss in stereo format?

Will this be like arma 2 were the amount of samples is capped and going over that cap stops the first of the sound files from playing.

Many regards & Thanks for the communication

Bp

in audio production, limiter affects the volume of anything that goes in, it cuts to loud peaks of audio, making it not to clip, (distorted) audio starts clipping at over 0dB, also you must not be afraid of damaging your highend audio system, by overcliping audio at large battles.

[-=seany=- 5]

Nice update thanks, looking forward to hearing them. Also, 60 % is a massive gain!

[NeuroFunker 11]

yup, now they can add, twice as much sounds as now, and it will stay about they same size, amazing!

[bigpickle 0]

in audio production, limiter affects the volume of anything that goes in, it cuts to loud peaks of audio, making it not to clip, (distorted) audio starts clipping at over 0dB, also you must not be afraid of damaging your highend audio system, by overcliping audio at large battles.

Yeah i understand what clipping is, all audio files should treated for clipping before being used or implemented in the game engine anyway.

Correct me if I'm wrong ofc, your suggesting that they are using code to treat clipping in game but more aimed towards "clipping caused by amounts of simultaneous files playing at once".

[Ezcoo 47]

Nice to know that the sound engine is being enhanced :) May I ask that will the sound positioning be enhanced later? That would be more than amazing (and desperately needed :p)

[mr.g-c 6]

I dont know where the benefits of stereo sounds are for a game?

Anyone care to explain?

[NeuroFunker 11]

Correct me if I'm wrong ofc, your suggesting that they are using code to treat clipping in game but more aimed towards "clipping caused by amounts of simultaneous files playing at once".

exactly that. No matter how polished the audio samples are, the game engine supports up to 128 samples, imagine what happens, if all of them playing at same time, there would be always clipping, until there is something, that doesn't allow it to clip, like software audio limiter.

[AnimalMother92 10]

Cool, always nice to see audio improvements.

[bigpickle 0]

exactly that. No matter how polished the audio samples are, the game engine supports up to 128 samples, imagine what happens, if all of them playing at same time, there would be always clipping, until there is something, that doesn't allow it to clip, like software audio limiter.

like a sound cap, like in A2 and OA to me, but I guess we'll see when its finished. Having tested the latest patch today I am sure there is lots more work to be done, its a shame really so much was broken and with missing sound classes from configs and so on it kinda seems like QC has gone awry.

[mikero 79]

Will we get a tool to support this or does it mean the engine now processes wss in stereo format?

The current wss format supports stereo. A few (not many) wss files in both arma and ofp were rendered that way (perhaps unintentionally). The engine ignored the 2nd channel, afaik, but the format, already supported stereo.

Most 3rd party wav-> wss convertors 'out there' already do this conversion for the current formats, plural. Providing those older formats remain supported in the shiny new wss, those tools will still do the job.

[bigpickle 0]

WSS converter gives a warning that the file is not a 16bit mono file, should i ignore it and crack on?

[ericfr 0]

Wind noise is too loud.

kind of cicada everywhere even in winter.

[mikero 79]

WSS converter gives a warning that the file is not a 16bit mono file, should i ignore it and crack on?

so far as I recall that exe would then not create 4 bit nibble compressed pch (if that was wanted). Secondly, perhaps now removed, the engine would flag such wss in it's (dot)rpt. But, again, so far as I recall, the resulting wss worked 'ok'

A better solution is wss2wav.exe (which does wav2wss too) available at

https://dev-heaven.net/projects/mikero-pbodll/files

There were additional 'undocumented features' with the bis version in producing 4 bit nibble pch. This format contains severely scrunched data which is excellent for engine noises and keeps the payload respectable, but atrocious for voice. The exe mentioned above produces 8 and 4 bit compressed stereo which the engine understands but the bis tool, does not.

As for "what's so special about stereo". It changes a tank coming at you from everywhere to a tank coming at you from the left.

[unknown03 2]

Awesome!

Does this mean that firefights wont sound like tapping pencils on a desk, or popcorn in the microwave?

[bigpickle 0]

so far as I recall that exe would then not create 4 bit nibble compressed pch (if that was wanted). Secondly, perhaps now removed, the engine would flag such wss in it's (dot)rpt. But, again, so far as I recall, the resulting wss worked 'ok'

A better solution is wss2wav.exe (which does wav2wss too) available at

https://dev-heaven.net/projects/mikero-pbodll/files

There were additional 'undocumented features' with the bis version in producing 4 bit nibble pch. This format contains severely scrunched data which is excellent for engine noises and keeps the payload respectable, but atrocious for voice. The exe mentioned above produces 8 and 4 bit compressed stereo which the engine understands but the bis tool, does not.

As for "what's so special about stereo". It changes a tank coming at you from everywhere to a tank coming at you from the left.

Excellent info and thanks for the link Mikero, I grabbed it but I get this error when i drop a wav file on to the exe, Ive checked my file is 16bit but i guess I'm doing something wrong.

Edited May 8, 2013 by Bigpickle

[sqb-sma 66]

I think some of the functionality from JSRS should really be in Arma. Most notably the change in how a gunshot sounds with distance, from bang to pop. With separation of mechanical and gunshot sounds we're off to a good start, but if the distant sounds could be put in it would be fantastic.

Apart from that my main complaint is in regards to bullet snaps, which sound like hollywood style "swooshes", once again JSRS, though awfully low quality (probably from youtube videos?), is perfectly on the money. Keep in mind even if you aren't close enough to hear the actual loud snap the bullet still produces a rattling sound as that snap is reflected off nearby objects (especially in alleyways/CQB).

[afp 1]

The sounds are pretty messed for the moment but looks like there is a nice change in the spatial effect and the way sounds mix. Looking forward for the full sound system.

@SMQ-SMA - bullets flying sound:

- Hollywood style may be close enough I guess...

[NeuroFunker 11]

I think some of the functionality from JSRS should really be in Arma. Most notably the change in how a gunshot sounds with distance, from bang to pop. With separation of mechanical and gunshot sounds we're off to a good start, but if the distant sounds could be put in it would be fantastic.

Apart from that my main complaint is in regards to bullet snaps, which sound like hollywood style "swooshes", once again JSRS, though awfully low quality (probably from youtube videos?), is perfectly on the money. Keep in mind even if you aren't close enough to hear the actual loud snap the bullet still produces a rattling sound as that snap is reflected off nearby objects (especially in alleyways/CQB).

yep i agree, especially, since JSRS 1.4 and 1.5 are perfectly working on arma 3 engine, in AiA mod. Distant sounds, reverberation on houses, hangars etc.

[gvse 10]

A lot of weapon sounds are missing, but they warned us about that.

Now, what I do not like in new sounds is the compression that's been applied to speech. When the characters talk in game the sound reminds me of a heavily compressed audio file. I don't know the right terminology to describe this, but I hope you see what I mean.

[fragmachine 12]

Thing that is bugging me in ArmA 3 sound engine is sound balance. For instance, we have a machinegunner firing from 100, 250, 500 meters 270 degree on asimuth (to our left). We can hear these shots on certain distance, certain volume etc in our L-channel - and then R-channel sounds (for example enviromental, or any other sounds that are much more silent) are almostly gone. In reality human ear is much more sensitive and it is adaptative to some degree. In my opinion there should be better balance for sound positioning and rest of enviromental sounds.

It would be nice if occlusion would be included in finished product and sound deformation/frequency echo and reverb - all affected by environment - that would take off a lota work with recording shitload of samples. Of course sound positioning is much more important, it's indicator of danger.

Just to mention some peoples out here on forums saying that "we don't need extra stuff like better sound engine, physics even graphics because it's tactical simulator not some stupid COD or BF" and then going to another topic replying to someones thread "dont expect assets from VBS2 - it's game not simulator fckin bullock" - well, I don't agree with both statements :)

As some of You may presume sound is important aspect for me, some parts of sound engineer still survived in me :)

[kamaradski 10]

Stereo sound samples ? Could someone explain to me how that works ?

What i thought so far was:

- Sound source is submitting a mono sound sample into gameworld

- For the one listening stereo is then created by hearing submitted sounds from different directionalities, right ?

Reallife example:

My car engine is submitting only one sound, if i listen with my right ear, it's in my right channel, if i listen from the left ear, it shifts to my left channel.

Yes standing next to my car i hear maybe left louder then right if i face the car left-sided, however in a game engine you could never configure that if the objects are not remaining static.

The only use i can see for stereo sound samples is to be able to transmit 2 channels of sound simultaniously and thus maybe saving some resources. However then limiting yourself with possible sound combinations.

Example: the bang of a exploding cardridge and the clicking of the gun mechanics. However in this case you could not re-use the gun mechanics sound in combination with a different bang for lets say fireing different ammo.

So i'm a bit clueless, unless the soundengine is not really spacious and directional it doesn't make sense to me ?