Opus Audio: Pairing Skype's SILC With CELT
Phoronix: Opus Audio: Pairing Skype's SILC With CELT
For those that haven't heard, the IETF Codec Working Group has paired the technology from Skype's SILC codec with the CELT codec from Xiph.Org to form the Opus Interactive Audio Codec. This new codec can be used for VoIP, live music streaming, and more...
My understanding is that Opus can produce better quality at lower bitrates than Vorbis for certain categories of audio, such as music, and even more for voice.
Since the MicroSD cards available for phones are pretty small, it'll be neat to get an Android player that can do Opus and re-encode my entire music collection in Opus -- at about 64kbps, which would sound terrible for Vorbis, but might be as tolerable as 128k MP3 in Opus.
Any VoIP app currently supporting Opus?
Any music player currently supporting Opus?
Any encoding app currently supporting Opus?
Where to find which apps do use the codec?
I just don't understand these projects that seem to be "floating in mid air", with no real world support. There should be a list at their website announcing the projects using their codec. I'm not saying "nobody's using this", I'm just amazed that the developers don't find it important to inform who's using their great work for real life usage.
It's still in development, so I imagine it's not currently considered production-quality. However, Mumble is pretty quick to adopt new codecs to improve quality and reduce latency (they were one of the first projects to use CELT), so I imagine they will add it as an option pretty soon (PLC is a huge feature in and of itself).
How does this compare with Skype's SILK codec and Xiph.Org's CELT codec?
Is this the best of both, or is it some crippled variant of SILK?
as usual Michael is behind the times with anything libav/ffmpeg related (virtually any AV related 3rd party app worth mentioning uses one or the other today as its core work horse) but basically any app that uses a current libav/ffmpeg can use opus decoding now once they update their version with this patch , im not aware of any encode patches at this time, not that it matters as the only one that matters is AAC along with AVC for video if you care about hardware assisted encode/decode and lower power usage not to mention better quality AV per used bit all round
Originally Posted by Aleve Sicofante
May 22, 2012
... This patch adds preliminary Opus audio decoding support with libopus for libav.The patch does not currently utilize multistreams, channel mapping and output gain.
Retaining copyright to Nicolas George (Cigaes) as most of the code was reused from the existing CELT decoding code."
Last edited by popper; 05-31-2012 at 02:03 PM.
I don't comment here on TMZ often, but I would like to offer my 3 cents for anyone who has made it to this post.
I don't code on the actual codec, but I have been hanging out and working with the people who do for over a year now. I'm a computer audio and music enthusiast, and have been listening to music stored in computer files since the mid 90's when hard drives were only slightly more roomy than CD's. The vast majority of my personal music collection is in FLAC format, and I am particularly annoyed by compression artifacts in all forms.
Opus is the best lossy audio codec ever developed. There is only two scenario's where you would not want to use Opus. The first is if you want a perfect copy, then you would use FLAC. The other is if your trying to use insanely low bitrates such as below 1K per second, in such case you want the codec known as Codec2.
In my experience Vorbis is 'transparent' to me (meaning no way I could tell you if it was lossless or not by just listening) at Q7, which is about 192k per second. MP3 at 320k, and for AAC I haven't heard any at a high bitrate so I can't tell you, but every use of it is at a lower rate, which sounds quite bad to me.
To me Opus is transparent at 128k, and this means for particularly hard music for codecs such as Norwegian Black Metal. Minimal techno and related music could fool me at 64k per second. Above around 96k per second you probably couldn't do a scientific study and get meaningful results, but I suppose well have to wait for the science to come in on that. The actual developers of the project maybe more conservative on sweeping statements and such, but I am not afraid to stick my neck out. For the vast majority of digital audio needs, Opus is the best choice.
The core software is pretty much ready to go, and we are working now on spreading support across the open source ecosystem, if you are a developer of audio software, stop by #opus on freenode IRC. We are happy to help.
As I understand it, the design goal was not so much higher quality, but mainly low delay. People really hate it when they say something and the computer has to take a long time to code their speech. I personally would never reencode my music collection to opus, and just leave the high bitrate vorbis.
Originally Posted by allquixotic
Originally Posted by Wilfred
SD/MMC cards are getting bigger and bigger.
No phone supports Opus yet, and when they do, memory cards have gotten bigger and cheaper.
P.S. Transcoding is a sin, and those who commit it will burn!