Voice chat codec suggestion


#1

I think you guys should use Opus Interactive Audio Codec for the voice chat codec

http://www.opus-codec.org/

Opus is a totally open, royalty-free, highly versatile audio codec. Opus is unmatched for interactive speech and music transmission over the Internet, but is also intended for storage and streaming applications. It is standardized by the Internet Engineering Task Force (IETF) as RFC 6716 which incorporated technology from Skype’s SILK codec and Xiph.Org’s CELT codec.


#2

Like i mentioned before, Speex is also an alternative. http://www.speex.org/

Both are open, i think both work well and are proven.


#3

I think if u can suggest a codec that lowers the latency without reducing the sound quality


#4

Speex is not being updating anymore it was stopped in favor of OPUS most of the people who worked on Speex and CELT codec have or are working on OPUS and The original stand-alone CELT has been merged into Opus.

Xiph.Org now considers Speex obsolete; its successor is the more modern Opus codec, which surpasses its performance in all areas.


#5

Opus would be a codec to do just that


#6

I think Speex works just fine. Not sure if much has to be changed about it. But like it said, they are both open, so i don’t mind either way.


#7

Opus is an excellent codec with a license compatible with HiFi’s opensource license of choice. It’s a very good all purpose codec allowing one to make choices impacting latency, quality and bandwidth. It certainly has potential to allow qualities at or so near levels indistinguishable from 24K S/S @ 16 bits uncompressed for a fraction of the uncompressed stream @ 768KB/s.

I converted some high data rate AIFF format files here - 100MB+ down to 10MB then analyzed each file via FFT spectrum analysis leading to a result of - only 1-2% of Earth’s population should be able to actually discern a difference in either format in a blind test.

Bonus points for Opus being well supported by VLC thus libVLC and IceCast with ogg packet encapsulation of Opus data.

Given some research into finding an encoding rate maximizing compression and minimizing latency it seems using it makes a lot of sense. Opus encode client’s output - decode at audio server - mix as needed - recode for streaming back to clients.

A potential complaint would be opus is not supported by DJ software, like, Virtual DJ. For that there’s a simple option - using libVLC - you already have access to traditional shoutcast/icecast streaming playback - so they could use the tools they’ve come to know and love, streaming exactly as we always have.


#8

I doubt i would be excited about this low fidelity concept, seems out of step with the project name.


#9

What’s low fidelity about using a codec that decreases size of stream data rate with no discernible reduction in sound quality?

Yes - most of the time we think of what SL voice or other highly compressed formats sound like. Absolute crap. But - that’s because they go for the dead minimums needed to convey speech. Allowing for more quality and less latency we might end up with voice streams with many times the 6K/s rates (approx) those use but still a fraction of 768K/s.


#10

Judas, seriously, i think what you are concerned about is the music and please trust me when i say that if you want you can setup a 320kbit shoutcast or icecast stream and you will hear no difference unless you are extremely gifted.


#11

Im really enjoying the sound in hifi, I would hate to loose what we have.
I believe this cascaded audio idea will be the new camping. paid to use your spare capacity sounds good to me


#12

Put a different way… can you compress images such that you make an acceptable trade off between some absolutely pristine uncompressed format taking up hundreds of megabytes per image and something that’s a fraction of while still looking great? Absolutely. Can you take it too far in interest of making smallest possible file size leading to a horrid looking image? Absolutely. Sound codecs are no different. You can make intelligent choices in specifying parameters to have a quality of sound that’s high and a data rate that’s lower.


#13

if you want to here how it sound look at these samples

http://opus-codec.org/examples/

samples demonstrate the quality achievable with Opus.


#14

I’m getting annoyed by you telling me what I can hear. and that We all dont have nice mics. We all use Senheiser headphones and blue mics, well from the people I asked :slight_smile:


#15

And i’m using AIAIAI TMA-1 Studio headphones and a shure microphone. Believe it or not, but i am a musician, too.

But other then audiophiles there was never any market interest in lossless quality music. Go and ask all the music retailers how often they sell wav or flac as opposed to mp3.

What i am getting annoyed with is the ease with which you seem to want to waste everyones bandwidth. And on voice communication no less.


#16

I know that audio is especially important to some - after going near 2 months almost deaf while waiting for new implanted tech to restore my hearing… I cried when I put on my Sony studio master headphones, loaded Virtual DJ and mixed 4 hours of trance just for myself.

All of my collection is in AIFF format or 320K MP3 if AIFF or WAV isn’t available. I now have hearing more in line with an above average TEENAGER - vs what I hoped to someday have again, that of an average 52 year old - and I am saying that with the right choices in codec you can have cake and eat it too.

For those wanting to do live performances with collaborators not all within a few mS ping time - the answer is time coded streams - yes - it’s complex, but, if all performers use NTP to sync local clocks and embed time code data into stream then - at the assembling end (audio mixer) it finds the participant with largest delay and syncs streams to MATCH giving ability to sound as if you’re all in the same physical location.

There’s no free lunch in all this - you can’t break laws of physics, well, not unless you have some spare quantum entangled inet links about (do you? > . <), but, using technology effectively - you can make it sound as if you can.


#17

My bandwidth is unlimited that’s becoming the norm so i’m not wasting anything. If your having bandwidth issues maybe turn off the audio server on your domain?


#18

There is no such thing as unlimited bandwidth. You’re just not using enough to piss your provider off.

And the professional server market is a completely different story altogether. Unlimited always means “Unless you are using too much”. Period. Otherwise Youtube would move all their servers to one of those neat unlimited offers.

It’s not becoming “the thing”. We have more bandwidth availible, it’s true, but that does not mean we can just put stress on the network at our leisure.

And in all of that, have you ever considered that there are users that are not from america or europe or japan or any other country that has awesome infrastructure ? Will we just ignore all of them aswell ?


#19

How do Netflix do it they broadcast hd video which i assume has 5.1 hd sound, they seem to do ok
Will we just ignore all of them aswell ?
Yes its next gen, hifi also wont play my old 78’s


#20

Netflix does it by literally owning server farms and paying big bucks on their connections.