Which is better in terms of quality, Opus or AAC?

That's not how codecs work. At what bitrate do you want to store your files? The answer will be either or depending on the question.

>>59007099
opus

>>59007132
publicly announcing that you're reporting a post is against the rules

>>59007099
they're not for the same purpose: vorbis is storage, opus is network streaming

apples to oranges

Does Opus officially store several dozen kilobytes of dead air? If you decode an opus file to WAV it will be a few hundred KB larger than the decoded FLAC source, and takes more bits to produce the same bitrate as other codecs. Is there a specific purpose for this? Otherwise I'd go with different codecs to store music. Vorbis is almost transparent at 160k and is good enough even smaller.

>>59007611
Why reply if you don't know what you're talking about? FLAC is the lossless archival format, not Vorbis.

>>59007722
>FLAC is the lossless archival format, not Vorbis.
right, vorbis is storage

>>59007741
You clearly don't know what you're talking about because you keep using the word "storage". Every file that isn't created in RAM and expelled upon reboot is stored. Care to elaborate on which specific storage you're talking about?

>>59007099
https://opus-codec.org/comparison/

At low bitrates, opus dominates. If you go into high bitrates, opus should not be worse than vorbis, which beat aac at high bitrates.

So pretty much, opus > aac.

>>59007811
oh shit

>>59007819
Except for in the ~48kb/s range in which opus is beat by HE-AAC.

>>59007722
Opus only supports 48KHz sampling frequency, and resamples any source that doesn't meet that.

If the source was a 44.1 flac, the decoded opus will be 48KHz, which should be bigger than the decoded 44.1 flac.

>>59007846
Source? (massive blind listening test link)

>>59007288
>publicly announcing that you're reporting a post is against the rules
Restarting your router isn't. And I'm not even sure why is he even reporting this thread.

>>59007811
You clearly don't know anything about rotational velocidensity, that's why I always save my music in BMP format.

>>59007900
>have to wait 120 seconds to continue shitposting

Also
>mods on /g/

>>59007741
OPUS is better suited for streaming, because it has lower codec latency (frames cover a smaller time window)... but it is also higher quality than vorbis, strongly so at low bitrates.

Vorbis is... legacy. That's all it is.

File: 04. Jeff Beck Group - You Shook Me.flac.wav.opus.png (3MB, 2064x2239px) Image search: [Google]

3MB, 2064x2239px

>>59007979
It's also more widely supported, smaller than Opus at the same bitrates because it doesn't resample to 48KHz for some dumb reason, and because of that doesn't take that many more bits to reach the same level of quality. If you spent all the money you had on super-ultra-high-fidelity studio headphones and don't have enough money to buy an extra SD card for your FLACs, then I'd consider using Opus over the smaller varieties of AAC or not-that-much-bigger Vorbis.

>>59007876
My ass, but look at these smudges.

>>59007979
Also, at bitrate where Vorbis and Opus start being competitive, that's when LC-AAC starts kicking ass.

>>59007611
>lossy
>storage

Y'all a bunch of autists, one can't say the difference between mp3 and flac in a blind test.

In terms of snake oil, you manage to beat /v/tards

>>59008257
At least Microsoft has recognized that WMA has been dead for many years, but why even have the bots anymore? I'm sure by now those who would be able to convinced to switch to Windows already would have.

>>59008122
Oooo you can really see the quality

>>59008324
I bet you have two inches thick wires with shielding connecting your computer to your faggot amp with that gay bulb

>>59008395
DELETE

File: 04. Jeff Beck Group - You Shook Me.flac.mp3.png (2MB, 2064x2239px) Image search: [Google]

2MB, 2064x2239px

>>59008331
To be fair here's MP3 at a similar bitrate.

>>59008122
Somebody fails to understand psychoacoustics, and its relation to lossy codecs.

>>59008444
AAC outright chops off everything above 16K but it's the best sounding out of all 3 at 128kb/s. What does that mean psychoacoustically?

>>59008476

That's an interesting claim. Can you link the tests pieces and the ABX results?

>>59007099
HE-AAC for bitrates below 64kbps. Opus for 64kbps and up.

>>59008543
You can't ABX "best sounding". It's psychoacoustics and not really measurable.

>>59008623
LC-AAC for bitrates 97-160k and Vorbis thereon.

Isnt opus said to be the best lossy audio compressor at literally any bitrate?

>>59007099
OPUS

>>59008707
>said
[by whom?]

>>59007099
FUCKING OPUS

>>59008724
The developers? I don't really know I was just saying

>>59008694
>You can't ABX "best sounding". It's psychoacoustics and not really measurable.

-_-

>>59008694
>You can't ABX "best sounding". It's psychoacoustics and not really measurable.
So you make a claim about audio quality, get asked about that claim, then immediately refute your own claim by saying it's not measurable.

You're a special kind of stupid.

>>59008694
they asked whether you can even differ between the three, and are not just imagining things, which is why they asked you to an ABX test

File: 1355628715201.gif (375KB, 174x138px) Image search: [Google]

375KB, 174x138px

>>59008122
>listening with your eyes

File: 1468050635397.gif (285KB, 300x168px) Image search: [Google]

285KB, 300x168px

>>59008122
>smaller than Opus at the same bitrates
>smaller
>at the same bitrate
uh, what?
the only size difference you can get between two codecs at the same bitrate is differing container overhead, and vorbis/opus both use ogg since they're made by the same people

FLAC > opus/aac

>>59013527
He means same target bitrate. Vorbis is a lot smaller than opus because it has a VBR algorithm that Opus lacks. (Its VBR mode is basically CBR)

Opus only has an advantage in bitrates under 128kbit. Encoding opus in 160-256kbit is a complete bloat/waste of space when you should be using Vorbis

>>59013796
FLAC and Opus are different codecs for different purposes.

File: FLAC will eventually replace MP3.png (243KB, 970x1088px) Image search: [Google]

243KB, 970x1088px

>>59013836
Opus was DOA, FLAC is more likely to replace mp3 at this point

>>59013849
Opus is becoming an industry standard. Nobody uses any lossless codec for voip, or video streaming.

>>59013869
Opus is not an industry standard.

>>59013875
Clearly you don't use voip.

Opus usually performs better and is the lossy code of choice when it comes to preserve quality.

>>59013816
>opus can't VBR
?

on a side note, does anyone know what causes 4chan to think some webm's contain an embedded file (when they don't)?
i can't figure out what's causing it and it's fucking annoying, i have to just fuck with it randomly until it works

fuck this
https://jii.moe/SkZU-1vKl.webm

If Opus had some reliable form of VBR, there wouldn't be any need for Vorbis, but then again it has the upscaling cancer so that's not likely.

>>59007099
>pick any recent music video on youtube
>check formats with youtube-dl
>"bestaudio" is consistently opus@160k
>most encodes are opus, some are vorbis (maybe compatibility reasons?)

Say what you want about youtube, but I'm sure they done their research. They want to save bandwidth and they want to preserve audio quality as much as possible.

>muh latency
it literally only matters for voip, even live streams can live with quite an amount of latency.

>>59013816
VBR constrains were removed years ago from the encoder
This is like saying that AAC lacks VBR since the FDK encoder lacks a VBR mode

>>59008623
Opus does well down to 8 kbps, in fact the one of it's encoders is shit past 64 kbps
Opus is king from 4 kbps until lossless territory

>>59008707
It is at any sane nitrate, but it is beaten at 2 kbps by encoders designed for such insanity
>>59013875
It is since its mandatory for WebRTC and will probably be part of NV1, YouTube already uses it on everything
>>59014042
Opus has reliable VBR and it beats Vorbis even in pure CBR
>>59014183
Vorbis was around before Opus, that's why it's still there
But Opus makes little sense at 160 kbps, at that bitrate Vorbis and AAC are pretty much equal
I goes Google is rolling it out on the top end before adopting it in the rest

>>59014329
opus reaches transparency before 160k from what i've seen, it's overkill
not to say it's a bad idea to overkill a bit, better a bit too much than a bit too little

>>59007099
opus

>>59013849
Not unless FLAC suddenly obtains low codec latency.

>>59013849
Opus and FLAC serve two different purposes.

>>59014042
> upscaling cancer

??

File: safari-firefox-chrome-opera-edge-1.jpg (26KB, 696x474px) Image search: [Google]

26KB, 696x474px

>>59007099
If you're developing a website that uses audio, I see it like this:

>Opus (96 kbps VBR) for Chrome, Firefox, Edge and Opera

>HE-AAC (96 kbps CVBR) for Safari

I think they both sound great for this bitrate.

>>59019443
Apple is a member of MPEG and refuse to implement standards.

They don't do Vulkan, or any recent OpenGL either, as they have their NIH syndrome "Metal" api they push instead, to ensure lock-in.

I just use Vorbis -q 7 when I want lossy encoding. 224kbps is low enough and I can be confident it's 100% transparent for everything. Opus is still too new to earn this trust.

Is Opus really the best audio like youtube-dl claims, or is it better to get a 44.1kHz format? Assume this is about music.

>>59021146
Yes. If you really want 44.1 you can resample it back.

>>59021146
lossy codecs completely destroy the original waves and replace them with a mess that bears little resemblance but perceptively sounds close enough. I'm surprised that doesn't annoy you, but the fact opus standarized at 48KHz does.

>>59020968
Why not AAC though?

>>59022624
>lossy codecs completely destroy the original waves and replace them with a mess that bears little resemblance but perceptively sounds close enough
That's a shit description.

>>59023947
AAC is covered by patents, and all programs that ship with it require a licence. Opus on the other hand, is free.

>>59013527
Read, faggot. Opus resamples to 48KHz, and even though it cuts off at 20KHz per channel, it keeps that extra space as dead air and takes up more bits doing it.

>>59023996
>AAC is covered by patents, and all programs that ship with it require a licence. Opus on the other hand, is free.

I actually meant why he chose Vorbis over AAC. He doesn't trust Opus yet, okay, but AAC is widely supported, is better than Vorbis, has earnt is trust and qaac is a very good encoder.

>>59024031
>it keeps that extra space as dead air
*facedesk*. You're aware opus is a lossy codec, right?

>>59024254
Read the goddamn thread or fuck off.

>>59023996
>That's a shit description.
Huh, how do you think lossy codecs do work?

>>59024274
Stop weaseling and answer... are you aware opus is lossy?

Just how do you think that does work.

>>59007099
OPUS BECAUSE IT RESPECTS YOUR FREEDOMS

>>59024326
OPUS because it wins the blind listening tests.

Everything else is bias.

>>59024360
Which blind listening tests?

>>59024429
https://opus-codec.org/comparison/

>>59024450
How come nowhere in these charts they say what they're testing for?

>>59007099
Their compression is very similar to my eyes. Both are superior to mp3.

>>59024521
Maybe click on the actual test links below, and read.

>>59024600
I did. They don't say what they were testing for, They explain the methodology and the results, but not the test itself.

>>59007855
>Sampling rates from 8 kHz (narrowband) to 48 kHz (fullband)

>>59024031
no it does not you illiterate faggot, maybe read about something first before talking shit

>>59008257
>/g/ - Technology

>muh resampling
I can't remember where I read it, but it was a conscious decision to do this because the benefit from having an optimised encoder for a single sample rate with high quality internal resampling was greater than the benefit of using the native sample rate.
That and most computer hardware works at 48k, so it was thought that audio was gonna be resampled anyway.

>>59008694
>You can't ABX "best sounding". It's psychoacoustics and not really measurable
you could've just said
>it's pseudoscientific bullshit placebo effect
but that just doesn't have the same ring to it, does it?

>>59013939
was that supposed to contain sound?
because /gif/ has webm sound, /g/ does not