Thread 106477371

Which bitrate do you prefer?

16

192 kbps seems to be optimal for music.

i don't care as long as it sounds good

>>106477371 (OP)
The only correct answer is the highest obtainable.

>>106477371 (OP)
96 kbps vbr is the most sensible choice for music, it's the sweetspot opus was intended for.
for voice 32 is already very good, 48 if you wanna go crazy and 24 if you want to go conservative (whatsapp and telegram both use 24 for their voice notes and it sounds more than good enough for that).
at 128 kbps and above, a good aac-lc encoder (fdk, qaac) is on par with opus while having higher compatibility, so it defeats the purpose of using opus in the first place
see: https://listening-test.coresv.net/results.htm
>>106477507
192 would already be transparent with lame CBR mp3, retarded audiophools that use opus at such needlessly high bitrates should ack.

>>106477523
...which is 200 kbps more than the highest bitrate mp3 allows, only 200 kbps less than the average non-brickwalled flac (while still being lossy of course), and only meant for 7+ channels surround sound
...why

>>106477552
> 192 would already be transparent
Self fulfilling prophesy. Continued exposure to lossy audio causes cumulative brain damage, until you loose the ability to hear fine details.

>>106477371 (OP)
Whatever I get with OuterTune's "high" preset (YT music client on f-droid/izzy) which I believe is around YT music medium preset, and it's roughly equivalent to a 192kbps mp3.

>>106477596
>and it's roughly equivalent to a 192kbps mp3.
no, it's actualy better than that
still to this day I have no idea why youtube decided to go with 128, they constantly re-encode shit with av1 and abysmal bitrates but they really like their audio to be top notch for some reason, for most <=480p videos the audio track is actually bigger than the video track at this point.
funny thing is, audiophools have no idea of how this shit works and so youtube still has the reputation of "having shit audio quality" anyways, despite that not being true for the past 10 years.
>inb4 but the source audio could be garbage
the official topic channel uploads are sourced from lossless like on every other streaming platform.

>>106477577
This nigga thinks they know more than sound engineers with ~30 years of experience. It's only relevant in the imaginary world your mind creates.

Here, learn something, you can watch the entire video, but this's the relevant part:
https://www.youtube.com/watch?v=YgEjI5PZa78&t=291

>>106477667
So basically I'm milking yt "good enough" quality for free? And most people wouldn't feel a diference if they paid for the premium tier? Lmao

>>106477371 (OP)
I prefer the original in FLAC. But that's not an option for a lot of music, so I want the best quality I can get. I NEVER trust Opus files though because 10 out of 10 times they come from youtube.

>>106477688
Pretty much 100% of people have suffered from brain damage. Its almost impossible to find an untainted ear. Personally, I no longer even listen to music. There is no even any point, we all are so brain-damaged at this point its mostly just hallucination anyway.

>>106477770
correct.
the 256 kbps aac/opus from the premium tier literally only exists as placebo effect or to attract audiophools/retards that don't know anything about lossy audio or modern audio codecs.
and the same is true for most other streaming services as well actually, spotify for example gives you 160 kbps vorbis with the free tier and 320k with the paid tier... 160 kbps vorbis is already perfectly transparent for 99.9% of the population and 99.9% of music, but most people will assume it's "mid quality" because the only audio codec they know about is mp3.
again, no idea why these streaming platforms that are enshittified to the brim aren't cheaping out more on one of their biggest expenses (bandwidth), and not gimping the audio quality of the free tiers to push more people into paying, especially when people will still presume the quality of the free tier is shit, might as well save some bandwidth and serve them actually shit audio at that point.

>>106477795
>>106477780
>I hate things that are popular albeit I don't have sound arguments to hate them
here, kiss each other

>>106477780
>I NEVER trust Opus files though because 10 out of 10 times they come from youtube
...which in the case of bootleg remixes and stuff is often the only original source that exists...
so you'd rather download some "source > yt opus/aac > mp3" transcode someone acquired through some sketchy online downloader instead? doesn't make any sense.

>>106477795
trvkedRAW
same thing with vision btw which is why i stopped opening my eyes too

>>106477552
>opus 1.1
it's not 2013 anymore, what about the newer versions?

FLAC but downmixed to 16/44.1khz
You won't hear more frequently than that

>>106477371 (OP)
220-256, higher than 200 is nearly indistinguishable from flac, the extra bit makes it virtually impossible to tell the difference.

>>106477916
no significant improvements have been made to the encoder when it comes to high bitrate CELT mode since then, only some small improvements were made for hybrid SILK/CELT encoding at lower bitrates (<48 kbps) afaik.
but that's not really a problem, it just shows that the reference encoder already achieved very high levels of maturity a decade ago, actually this is a very good thing as you don't have an aac/mp3 situation where there are a dozen different encoders of very different quality to keep track of, if there's no encoding information in the metadata of a 128 kbps vbr AAC-LC file, you have no idea if it was encoded with FDK/QAAC (perfectly transparent) or (very likely) something like ffmpeg's AAC encoder which is actually even worse than lame mp3.
every single opus file out there has been encoded with libopus (99% chance it's at least 1.1 since it came out very soon after the codec released, no one was using it in the 1.0 days), which is the best efficiency the codec can offer, period, there's no reason to make/use a worse opus encoder (the reference implementation is completely free and open source) and it's tough to improve on it because it's already impressively good.

>>106477929
>higher than 200 is nearly indistinguishable from flac
let's see those 160-192 kbps abx tests you've performed then, retard.

>>106477371 (OP)
no 44.1kHz support in opus so for my music, I'll stick with AAC or Vorbis

Should I encode opus in 96 instead of 128? It's for listening to TLMC on my phone.

144kbps

>>106478299
>or
no bitch, you gotta choose

>>106477552
>96 kbps vbr
I definitely wouldn't trust opus to be transparent at that bitrate. I use 128kbps for opus and aac and 160kbps for vorbis.

>>106478344
https://listening-test.coresv.net/results.htm
opus 96 is really good

>>106477371 (OP)
96kbps

>>106477507
this is what I run. I know it's overkill, but it's already way smaller than V0 mp3 or FLAC

>>106477577
>until you loose the ability
Looks like you've been continuously exposed to lossy audio

>>106478299
>he's still posting this retardation
https://wiki.xiph.org/OpusFAQ#But_won't_the_resampler_hurt_the_quality?_Isn't_it_better_to_use_44.1_kHz_directly?

>>106478329
>Should I encode opus in 96 instead of 128?
yes
>It's for listening to TLMC on my phone.
96 kbps might even be overkill if you mean "listening with low quality earbuds while in public", 64 is already quite good for that use case.

>>106478344
>I use 128kbps for opus and aac
I definitely wouldn't trust a retard that doesn't even know how big the difference across aac encoders is.
128 kbps ffmpeg aac is like lame mp3 96 kbps cbr level of quality, you need to specify which aac encoder you're using for this to make any sense.

Whatever the default ffmpeg encoder uses. I'm not creating lossy encodes to archive lmao. Oh it's only 96kbps? Fine enough for my phone or whatever.

>>106478570
>way smaller than V0 mp3
not really
>or FLAC
correct, just like 192 kbps aac-lc with a good encoder or even vorbis would be.
such bitrates simply make the use of opus pointless as you're trading off compatibility for no gain.

>>106478626
>Whatever the default ffmpeg encoder uses
which is a very sensible default as ffmpeg devs aren't complete retards like half the anons in this thread are.
>Oh it's only 96kbps?
that's what opus needs to reach transparency levels for most music, you seem to imply it's little when it's not.

>>106478594
listening with $20 chinese iems of course

>>106478111
human perceptional differences are different fomr objective ones, maroon

>>106477371 (OP)
>voice
32
>mono video audio
64
>stereo video audio
128
>surround video audio
256
>music
200+, or whatever most converters have as the highest quality setting

>>106478830
what's with these "highest quality setting possible" retards here
you started off good, then went to complete retardation
>256 for surround but 512 for stereo music
what the fuck?

>>106478805
>nonsense word salad
ok

>>106478864
>illiterate retard
can't say I'm surprised

>>106478853
most gui encoders usually stick the "highest quality" vbr mode for stereo at about 190-220. i've noticed too many compression artifacts at 96-128 with music over headphones but it's fine for general audio over speakers

>>106478615
>128 kbps ffmpeg aac is like lame mp3 96 kbps cbr level of quality
That used to be the case (which is worse than not shipping an aac encoder at all imo), but I'm pretty sure the new one is decent enough.

96Kbps because all my headphones are trash

>>106478111
I have been part of a well known choir, I have objectively nearly perfect hearing, and I can barely tell the difference at 200kbps with a pair of decent monitors using a special sample set, at 220 its impossible for me to tell. maybe there is some god-like ubermensch that can do better, but I can tell you for 99.9% of audio enthusiasts, they will be unable to spot the difference in a blind test

I will also note that opus doesn't sound "bad" until below ~110-140 for me depending on the audio, which is impressive.

>>106477371 (OP)
160-185 out of old habits with Vorbis, even though I know deep down that 110-135 would sound about the same.

>>106478914
Are you talking about mp3?

>>106477371 (OP)
I use FLAC and don't care about bitrate.

I have some 24/96 7.1 tracks which reach 7-8000 kbps.

>>106479561
what tracks are 7.1 even

>>106477371 (OP)
PCM is the only option, im not a nigger

>>106477507
yeah i know i could be using less and it would be more or less identical but my autism wont allow me to go any lower than 192

>>106478914
>i've noticed too many compression artifacts at 96-128 with music over headphones
no abx test provided and never willl be, as always.

>>106479050
>and I can barely tell the difference at 200kbps with a pair of decent monitors
another retard that's not going to provide any actual abx test result
wikipedia.org/wiki/Placebo
when are you going to learn

>>106478921
>That used to be the case (which is worse than not shipping an aac encoder at all imo), but I'm pretty sure the new one is decent enough.
source? couldn't find anything about it.

>>106478329
96 or lower. Hell, I use 96kbps Vorbis for my phone and I could probably go lower because I only use earphones.

>>106477371 (OP)
I personally use 160
good compromise between 96 and 192 fags arguing with each other

>>106477371 (OP)
Full.

Only poor people who can't afford proper audio equipment are ok with lossy compression.

>>106482167
>implying the limiting factor isn't the human ear
yet another braindead audiophool
the most expensive equipment in the universe still isn't helping you abx shit.

>>106482332
Quality audio equipment massively exposes the mess in lossy compression. You can't discern that whith generic garbage so everything sounds the same. CD quality's good enough, and I ain't taking anything less.

>>106482408
sure, let's see those abx test results then

>>106482427
And how am I even going to certify my test? This is retarded.

>>106482496
correct, you could absolutely falsify the test by using 6 kbps or some shit like that while claiming it was 320 kbps...
and despite that, none of you audiophools ever bothered to even attempt that, hilarious.
you could perform some of the https://abx.digitalfeed.net/list.html tests, maybe there's a way to cheat here too but probably not that straightforward to do.
but that doesn't even matter because again, you aren't even going to try, I always ask dozens of audiophools in each of these threads and it always ends in *crickets*.

I've read somewhere that 160kbps is transparent, so I just use that

>>106482818
transparency depends on your own hearing

Considering I'm an audiophile with an extremely high fidelity / high resolution system, I only listen to 16bit 44.1khz FLAC files.
Anything over that is mostly a scam and anything less than that is a noticeable loss.

-c:a libopus -b 192k.

I should start saving shit with a bitrate of FIVE TRILLION desu.

>>106478344
it is for 99% of use cases

https://opus-bitrates.anthum.com/

>>106483898
i encode webms for /gif/ with 32kbps only, it has always been enough so far.

>>106482637
Haven't you considered the possibility it's your retarded attitude what's actually keeping people from wasting their time with your tests?

>>106477507
Im fine with that for DJing.

>>106483196
Wisest of men.

>>106483955
Also, bitrates become relative once you have enough LSD or similar stuff in your blood.

No 128kbps though. Nor lower. 160 is the minimum.

>>106483942
>nonsense excuse to not perform tests
classic

>>106484178
What's in for me?

>>106477371 (OP)
Encoding music into opus adds delays before playback with seemingly completely random tracks. I use fre:ac, then when I open the file in foobar, it can hang for a whole minute before the track starts playing. Applies for both the desktop and mobile versions. VLC, Audacity plays such tracks fine. What the fuck?

>>106484192
...proving you're not a complete fucking retard that doesn't understand the very simple concept of placebo effect?
this is how anything in life works, you have to provide proof to claim something, do I even have to explain this?

>>106484212
whatever dogshit software you're using to encode is likely the culprit.
the encoder doesn't add any delay and also works perfectly fine for gapless playback, such an obvious flaw would instantly make any bitrate option non-transparent.

>>106484229
Like I said, I'm using fre:ac, which as far as I know is standard, unless you want to suggest something better.
The issue occurs with only something like 1% of tracks, with varying severity. Sometimes the delay is only one second, sometimes a minute.

>>106484214
Nope. It's either properly certified somehow or I ain't doing shit, because first thing you're gonna say is "fake, you cheated". Eat ass and die.

>>106484238
>as far as I know is standard
never heard of it
>unless you want to suggest something better.
foobar2000, which you already have.

>>106484240
so you can't even read, nice.
I literally told you to use https://abx.digitalfeed.net/list.html
>because first thing you're gonna say is "fake, you cheated"
I already explained to you that there's no point in worrying about that, because you aren't ever going to perform any ab test in the first place.

>>106484258
If I can't download it directly, I can't play it back on the stereo. I don't have an expensive interface or monitors on my computer. In any event, not certification, no shot.

>>106484325
>his trillion dollar poorfag-mogging audiophool grade stereo system doesn't even have an aux port or any way of connecting to any external device
see? i was right, you were never going to do an abx test in the first place, what a retarded excuse.

>>106484410
>his trillion dollar poorfag-mogging audiophool grade stereo system doesn't even have an aux port or any way of connecting to any external device.
If I have to play it from my computer, it's already messed up because the sound card is ass. You shitskins are something else.

>>106477371 (OP)
96kbps is good enough for my phone use, I go outside it's loud as fuck between all the cars, trains and retards and my $1 chinkbuds probably can't perform any better
at home I found it that above 128kbps I could no longer hear any difference but then if I'm at home I'm listening to FLAC anyway, not because I can hear it but because why not at this point

>>106484448
>he can afford a billion dollar sound system but cannot buy an 8$ apple dongle
you still aren't making any sense, a perfectly transparent usb sound card costs literally pennies, it's very likely that the one in your computer is already perfectly transparent but since you're an audiophool and it costed pretty much 0$ you'll just imagine it sounds like shit.

>>106484539
>a perfectly transparent usb sound card costs literally pennies,
That's the thing, you have no idea how bad it actually is till you're able to compare with something good, and even if it weren't so, I don't work in the audio industry and I don't do anything audio-related on my pc which may even partially merit getting a better setup (and I ain't ordering one just to shut you up). I have my high-end stero where I play all my cd's, flacs and vinyls and that's about it.

>>106484613
>That's the thing, you have no idea how bad it actually is till you're able to compare with something good
my fucking sides
ok, that's it, complete total audiophool confirmed
no point in continuing this discussion then, my bad.

The human eye can't see more than 96kbps.

>>106481515
you ask for evidence, why should I do anything , when you do nothing yourself. I spent like 3 hours tweaking a file for size while retaining as much detail as possible, thats what I'm basing my anecdote on. you can also read the ffmpeg encoder page for basically the same conclusion, I basically wasted 3 hours to confirm exactly what that page says about quality.

inb4 "burden of proof", all I can say is that this is a Nigerian basket weaving forum, and your expectations are way too high.

I'm using 320kbps hybrid wavpack and set up syncthing to only send the lossy part to my phone. If I run out of space again I'll probably make a mirror folder with all flacs converted to 160kbps AAC.

>>106485860
Why