[VoiDeD]

Have you tested using an instance of ISteamUser from the gameserver's context rather than just creating a new steam user instance inside your extension? I also find the usage of IClient interfaces suspect as that'll be prone to breaking randomly when valve decides to change the interfaces (and that OSW generally doesn't keep these up to date).

[Zephyrus]

Originally Posted by VoiDeD Have you tested using an instance of ISteamUser from the gameserver's context rather than just creating a new steam user instance inside your extension? I also find the usage of IClient interfaces suspect as that'll be prone to breaking randomly when valve decides to change the interfaces (and that OSW generally doesn't keep these up to date).

i have, unless your account type is set to individual the decompressvoice function prints a failed assert and refuses to work. also as far as i remember most games dont have IClientAudio at all anymore, I think at the time I wrote this only a single game used that still.

also i fixed the missing files in the src zip now

[robotortoise]

Originally Posted by Zephyrus When I wrote this monthes ago I streamed the decompressed voice data through websocket to a website and it was possible to listen to the ingame voice chat

That's actually really neat! I wonder if you could stream users' voices to mumble and the responding voices back to TF2, so the people in game could communicate with people in the mumble without actually using mumble. It'd allow for higher quality voices, at least for the people using mumble, without sacrificing ease of use for the game players.
Maybe a !mumble command to activate it or something, so mic spammers wouldn't be broadcast to the mumble.

Am....am I making sense here? I can try to rephrase it if you'd like...

[Zephyrus]

Originally Posted by robotortoise That's actually really neat! I wonder if you could stream users' voices to mumble and the responding voices back to TF2, so the people in game could communicate with people in the mumble without actually using mumble. It'd allow for higher quality voices, at least for the people using mumble, without sacrificing ease of use for the game players. Maybe a !mumble command to activate it or something, so mic spammers wouldn't be broadcast to the mumble. Am....am I making sense here? I can try to rephrase it if you'd like...

its probably possible to stream voice from server using mumble-ruby but not sure if its worth it at all..

[robotortoise]

Originally Posted by Zephyrus its probably possible to stream voice from server using mumble-ruby but not sure if its worth it at all..

It'd be useful for my server and other mic spam servers at least, because we have a separate mumble for talking and music playing over the TF2 in game chat, so when people try and talk over the mic spam they can't be heard.

[Zephyrus]

Originally Posted by robotortoise It'd be useful for my server and other mic spam servers at least, because we have a separate mumble for talking and music playing over the TF2 in game chat, so when people try and talk over the mic spam they can't be heard.

well first of all to make this more or less usable you would have to write some kind of audio decoder for javascript (although probably possible with emscripten?), cause raw pcm wouldnt be a viable solution with many players

[robotortoise]

Originally Posted by Zephyrus well first of all to make this more or less usable you would have to write some kind of audio decoder for javascript (although probably possible with emscripten?), cause raw pcm wouldnt be a viable solution with many players

So, uh....that's difficult, right?

[Zephyrus]

Originally Posted by robotortoise So, uh....that's difficult, right?

well as far as i can see mumble-ruby has the ability to read raw pcm from a FIFO and play it back on the mumbler server. so that way is probably easily possible. to do it the other way around you would have to open a hidden motd for the clients with a javascript websocket client that receives audio data for playback from the mumbler bot. but raw pcm audio would be a waste of bandwith and wouldnt even be possible with many servers. therefore you would need some codec (like speex as its pretty simple) that you compile for javascript using emscripten so you can decode incoming speex audio from the mumbler server. or you maybe you could even use mumbler's opus codec if you get it to compile for javascript, its not that trivial though

[Peace-Maker]

Can't you inject other voice data into SV_BroadcastVoice and make e.g. a bot talk what ever comes from mumble or anything? Let the server do the networking and encoding and the client use the default embedded voice decoding, so you don't have to rely on some browser hacks?

[Zephyrus]

Originally Posted by Peace-Maker Can't you inject other voice data into SV_BroadcastVoice and make e.g. a bot talk what ever comes from mumble or anything? Let the server do the networking and encoding and the client use the default embedded voice decoding, so you don't have to rely on some browser hacks?

Well it would take up a slot, but yes it's possible