Slashdot Mirror
Opus 1.2 Released
jmv writes: The Opus audio codec, used in WebRTC and now included in all major web browsers, gets another major upgrade with the release of version 1.2. This release brings quality improvements to both speech and music, while remaining fully compatible with RFC 6716. There are also optimizations, new options, as well as many bug fixes. This Opus 1.2 demo describes a few of the upgrades that users and implementers will care about the most. It includes audio samples comparing to previous versions of the codec, as well as speed comparisons for x86 and ARM.
The 32/48/64 kb/s demos with 1.2 are astoundingly good for that low of a bit-rate. Nice job by the Opus team to get things sounding this good. Was also glad to see more robust fuzzing tests to help with potential security issues.
This is super-geeky stuff, but since I've integrated some of their older Ogg Vorbis stuff into previous game engines, I like to keep up with what they're doing. I might switch my game engine's decoder over from Vorbis to Opus at some point, but I've got to stop futzing with the engine and get my game out the door.
Irony: Agile development has too much intertia to be abandoned now.
Congratulations to the Opus team and xiph.org for doing such a great job with this. It's really a wonder that, with examples of F/OSS software like this, it's not simply the standard by popularity. I'm involved with an Android-based audio livestreaming app for Icecast (called Cool Mic) which uses Opus as well as Vorbis. My experiences with those involved with any of the Xiph projects have been great. Extreme talent over there.
It is pitch black. You are likely to be eaten by a grue.
I put it the other way: the more people use god-awful earbuds in noisy coffee shops the more we can reduce the bitrate without anybody noticing :-)
Opus: the Swiss army knife of audio codec
1. Someone (e.g. on the mailing list or on the Hydrogenaudio forum) points out a music sample where Opus performs worse than other codecs or just worse than it usually does.
2. We investigate to find out what's causing the artefacts and (especially) why this particular sample is affected.
3. We come up with a possible fix that improves the quality of that sample, without making other samples worse.
4. We look for other samples with the same characteristics found in 2. If the fix also improves them, then we go to 5, otherwise we go back to 3 (or sometimes to 2). In case of infinite loop, do some throttling (i.e. drop the issue and go back to it later).
5. When we're happy that we have an improvement, we clean it up, make it as general as possible, test it, and merge it.
They improved the codec by listening to user feedback. THAT'S CHEATING! Everyone knows that you need be a corporation that develops something decent and then you ignore all comments that aren't about how great it is and then ban the unhappy riff-raff! Do you think we would have ended up with smartphones so thin that they bend and/or explode corporations had listened to all the people screaming for a higher battery capacity?! Do you think we would have been graced with Thunderbolt ports with the same connectors as USB-C if they gave a damn about what people thought?! Do you think we would have overpriced and under-performing wireless earbuds and no headphone jack if they listened to even a single user?! OF COURSE NOT!
Where do you jackasses get off developing things that people want and like? And if you do, you aren't even going to demand they fork over their life savings for it?! OPEN SOURCE IS RUINING EVERYTHING! ;)
Anons need not reply. Questions end with a question mark.