What bitrate do I use to get close enough to FLAC quality that there's no audible difference?Same question for voice-only audio books and podcasts.Is 32 kbps enough for everything?
>>10318745196 is enough for using on the phone
>>103187451Try taking an audio file and converting the source file to multiple Opus files at different bitrates to test and see which one personally works for you.
Human ear cannot hear more than 160kbps
how do you get your audiobooks?
>>103187475I would but I'd honestly trust an autist with spectrographic tools and spreadsheets of comparisons more than I trust my own ears.
>>103187496I usually convert them from ebook with piper TTS, which outputs to WAV.
>>103187508What if autists want 64-bit pcm 192khz?
>>10318745124kbps per channel in Opus can hit 4/5 MOS now but unless you're really desperate for space it's better to stick to 64kbps per channel instead, just to be safe. It's enough of an overhead that you'll probably never look back and think, "dam I should have used more bits!".https://files.catbox.moe/aq2xqi.opus
192 for music64 for voice only
>>103187524Well you can usually tell when someone's autism has effectively made them retarded.>>103187626That's amazing but what is MOS?
>>103187713Mean Opinion Score tests used with a lot of people involved. 4/5 is a great score, think of it like giving an Amazon product 4/5 stars in a review.
>>103187451Pic related are the recommended settings by the developers of opus.https://wiki.xiph.org/Opus_Recommended_Settings96 - 128 KB/s for Stereo and it is transparent.I round up and use 160 which puts the file size of an average song at 3 - 4 MB. It's as large as a very-lossy mp3 while definitely sounding like a flac.
>>103187753This is a dumb question, but are the listed bitrates "per-channel", as in 64 + 64 = a 128 kbps file? And when audio players list kbps is it usually per channel or total? This shit is confusing.
>>103187753Pretty outdated given all the improvements to Opus in the last few years. Probably not worth re-doing thousands of MOS tests just for 20-40% compression efficiency gains though.>>103187781If you don't see the word "per channel" it usually refers to stereo but IMHO people really should have left it at "per channel" since not all of us are stereo plebs.
>>103187520thanks
>>103187451160 kbps for music32 kbps for speech
>>103187475Not a good idea, because it depends on the gear.Maybe at some point in the future OP gets headphones with good treble and suddenly be able to hear a difference that he wouldn't hear now.OPUS acts like a lowpass filter and cuts of high frequencies first.
>>103187795>Pretty outdated given all the improvements to Opus in the last few years.there have been no improvements in audio quality for high bitrates since 1.3
>>103187795Per-channel is silly, because a codec can simply derive one channel from the other.Like save the second channel as a simple difference of the first one.The recommendation for 6 channels (256 kB) is only twice as large as for 2 channels (128 kB).
>>103187781>when audio players list kbps is it usually per channel or totalit's total
I just converted an audiobook as a test of low bitrates.>mono>2 hours length>6kbps>Total output size: 5 MBIt's old telephone quality but completely listenable. Very impressive.
>>103187865Dunno about "high bitrates" but there have been for low bitrates. Unless you have expensive DACs/headphones, 24kbps per channel OPUS will in fact score 4/5 MOS. Not always but enough to impress the shit out of most audio nerds no doubt. You can do the long ABX MOS testing yourself if you don't believe me.https://hydrogenaud.io/index.php/topic,113985.0.htmlPersonally I don't care, 64kbps per channel is the default norm now giving us all enough overhead to not really worry about audio sounding like muddied dogshit audio like we did with MP3. It would have been hilarious if instead of sucking mpeg-la cock all the time humanity banned together and started selling "Opus players" instead of "MP3 players" just to make these jews writhe in pain.
>>103187465No, it isn't, unless you use some shitty bluetooth headphones. Even 130kbps or whatever youtube uses is quite shit and nowhere near 256/320kbps MP3.
>>103187781No
>>103187451Its 2024, there is no reason for lossy to still exist. Anyone who willfully chooses to encode lossy audio should be skinned alive.
>>103189167I dislike bloat and think we should stop putting ourselves on a pedestal. Once you understand the biological limitations of humans you'll start to notice that certain things become acceptable such as what you call "lossy slop".Also not all of us are looking forward to having a closet full of HDDs. Some of us just prefer a small pile of USB flash drives instead.
>>103189353Lossy audio harms you. Its been improved to remove the obvious tells, but it still detracts in a non-obvious but real way. Also, millions of pieces of music have been lost to the void because of lossy audio, software capable of performing lossy compression should be illegal to possess.
>>103187700This. 192 CBR Opus is the way.
>>103189167To what end will lossless fags treat extensions as codec containers?
generally speaking, 128kbps VBR is enough for most cases
>>103190125>CBRNo.
>>103187520Holy shit that's niggerlicious
>>103187781Bitrate is the number of bits in a second per stream. The number of channels is simply the number of streams, it has nothing to do with bitrate.
160 is good for hi-end hardwave128 is transparent to normal people96 has minor noticeable artifacts64 is transparent if you're 40+48 has tolerable sound/quality ratio24 is only usable for speech. it's sufficable>48 is good for audiobooks & streaming radio>above 160 is placebo and only used by turbo-autists>if quality is non-negotiable, use flac, not lossy codecs
>>10318745196k Opus is pretty much equal to 320k MP3 to my ear, meaning that I don't hear the difference, unless I turn up the volume extremely high and autistically listen to very tiny artifacts for minutes. To me, that's transparency. Over 128k is apparently actually transparent, though I've never tried it since 96k is already great.>voice-only audio books and podcastsI have a bunch encoded to 64k and they're good.
>>103189393>millions of pieces of music have been lost to the void because of lossy audioIt's only "lost" to audiopedophiles which are in 99% of the cases retarded buyfags
sound test: https://opus-bitrates.anthum.com/
>try to convert a bunch of pirated FLAC music to opus>ffmpeg can't embed images in ogg/opus as a thumbnail>opusenc also seems to dick up some metadatapain in my fuckin ass>>103189167You are mentally ill.>>103191957Enough for music cases, very overkill for speech.
>>103187451Anything over 192kbps and it becomes indistinguishable to most ears.
>>103195596>128 is transparent to normal people>64 is transparent if you're 40+?
>>103187451>Is 32 kbps enough for everything?It's enough for stereo to not be distracting in a 26-minute video posted on a Tuvan yurt erection forum, but it's not enough to play on your Hi-fi loudspeakers as you lie down after consuming psychedelic fungi to guarantee a good trip.
96 for music32-48 for audio books and podcastsany higher than that and it becomes to pointless to use opus instead of aac-lc with a good encoder
>>103190125>CBRwhat the fuck why?
>>103198129my script prefers opusenc, though i haven't seen either fuck up the output. what metadata do you see opusenc fuck up?
>>103200092too lazy to check because this was several months ago but I remember there being something about the music-related metadata before & after. maybe I'm misremembering but there was a reason I stopped looking at it.
>>103187978I did some more tests with the same TTS audio.>1 hour, 24 kbps = 10.8 MB>1 hour, 32 kbps = 14.4 MB>1 hour, 48 kbps = 21.7 MBDoing the math for a 10 hour book:>10 hours, 24 kbps = 108 MB>10 hours, 32 kbps = 144 MB>10 hours, 48 kbps = 217 MBThe S's are slightly harsher at 24 and 32, but the difference between 32 and 48 is only barely audible. Honestly there's not a big enough size difference for me to go lower than 48 kbps for my audio books, but as in my previous test you could get 10 hours in as little as 25 MB at 6 kbps while still getting fully listenable speech, wich is insane.
>>103187451>close enough to FLAC qualityNothing. FLAC is lossless.>that there's no audible differenceI could not ABX a single one of my "killer samples" at 192 kbps. And only a handful of them at 160 - you actually have to listen critically with good headphones and even then - it's a handful of killer samples.If you're encoding shit to listen to in the car or on the bus or some shit you wouldn't be able to ABX anything at 128k.Just use qaac at V100. I've yet to encounter anything I could ABX. And AAC/M4A is more compatible than opus with dumber old shit.
notice how OP didn't post format war bait and how the quality of the thread is significantly higher than usual
>>103201309>close enough to flac quality>nothing>I could not abx a single one of my killer samples at 192 kbpsnice contradiction there pal
>>103201309>AAC/M4A is more compatible than opusI mean, yes, it goes without saying that if you care about compatibility, or don't care about trimming the filesize down, you don't go for opus
>>103203486Nothing lossy is going to be "close enough" to being lossless to replace it. I kept the FLAC on my computer and made qaac V100 copies for my phone because there is no audible difference to my ears.
>>103203704"good enough" depends entirely on your use case.
>>103203704>contradicts once againwe are talking about human perception here, retardand btw, stuff like 512 kbps vorbis is pretty close to lossless, it's just pointless
>>103203721If that use case is archiving opus doesn't cut it. For almost everything else it's excellent.
>>103203721Thank you, I am aware>>103203736I knew I was wrong to remove the "retard" from the first reply, you contrarian mouthbreather. It is not a REPLACEMENT. The lossless file is a REPLACEMENT for the original - because nothing is lost.>512k VorbisFucking moron.
>>103203749A replacement for all use cases is usually dubbed drop-in. 'Replacement' is much less pedantic than you think it is. You sound rather angered.
>>103203393because OP supports trans rights
>>103195918i couldn't tell the difference between 64kbps and higher bitrates with my headphones. 32kbps had obvious bitrate culling effects.
>>103205129same
bymp
>>103203393OP here. Tbh I did bait with the "Is 32 kbps enough for everything?" but I was just hoping to trigger informative replies.>>103204425Nope. KYS.
>>103205129same here
>>103189167I have my music collection stored as highest quality lossless FLACs I can find on my PC with expandable storage, but my phone with a fixed ~230 GB gets 160k Opus.
>>103198129My solution was to use ffmpeg to encode (opusenc will read replaygain tags and put them into header gain, which I do not want), then extract cover art using ffmpeg and embed it into the opus file using opustags.
>>103187451Just use flac.What on earth is wrong with you?
>>103188823lmao.I thought like you once, and then did a blind test on myself. It was a humbling experience.>>103187451I doubt anyone would notice 96kbps, and certainly not 128.
why are you people trying to save a couple of megabytes like this? do FLAC or mp3 V0 and call it a day. Even your phone has dozens of GB of storage nowadays.
>>103209083Not everyone likes bloat. That's like arguing in favor of mpeg-4 video codec over AV1.
>>103187451If you don't notice the difference, then it's good enough for you.Anyone saying otherwise is an audiopedophile./thread
>>103187451There's no difference between v0 and FLAC. >b-b-b-b you need to spend more money!!1!!!Shut the fuck up. Blind tests prove you're a fucking faggot who enjoys other men pulsing inside your asshole.
>>103208901It isn't that easy to hear the straight diffference between the codecs and bitrates on short A/B comparison, but there is definitive difference in general mood when listening to mainly 320kbps MP3 / FLAC files from local storage than Youtube Music etc. And I usually can definitely hear the difference instantly with certain (synth) sounds that have high upper harmonic content. It discards something that isn't easy to describe and regenerates something very generic in it's place.
>download one album in flac format>takes up fucking 1gb of your precious storage
>>103210048>there is definitive difference in general mood when listening to mainly 320kbps MP3 / FLAC files from local storage than Youtube Music etcThis could easily be caused by the Youtube Music and the FLAC editions being mixed differently, and not from a difference in compression. Do you notice a difference if you take one of your FLACs and compress it to 96kbps opus?
>>103209251>audiopedophileThis.
>>103209251Lossy audio causes cumulative brain damage. You know how depression and mental illness is rising to new heights, lossy audio is a significant contributor.
>>103213098meds now
>>103210060>16000gb can be had for $200>degrading your audio experience to save 1 cents worth of storage space
>>103209083While I can see the appeal of having a special minimal format that is good enough for your needs, incompatibility is not something I want to ever think about, so I stick to MP3 V0 for lossy.
>>103213512now add another one as a local backup, that's already 400$then when one of the hard drive fails? 200$ to replaceand off-site backup? no cloud provider has that much capacity unless you pay literally hundreds a monthso what's the point of storing 16 terabytes of flacs if you are going to lose all of it? retardand also... >degrading your audio experienceshow abx test result
>>103214118> +backup = 2 cents per album total>drive fails >>> it is now the future and I can purchase a larger, cheaper driveEven lossy homos admit you should archive losslessly anyway so all these costs apply to anyone who wants to store their own music files. Worst case scenario you lose it all and have to get the files from somewhere else again. Off-site backup is stupid because they are not your personal home videos, they are files that many people already have backed up. Rare stuff you can treat differently.>show abxShan't. The difference is palpable. If you haven't spent hundreds of dollars on headphones, DACs and amps you shouldn't even be allowed to talk about audio. Even then some people clearly have mental defects so the opinions of most reditors, for example, should be completely ignored.
>>103210048>difference in general moodFuck of with the audiophoolery. If you can't pass an ABX test, whatever difference you're imagining isn't real.>MP3LMAO. For many killer samples, 320kbps MP3 is worse than 128kbps Opus.
>>103214247>you should archive lossless anywaycorrect, and a single 16 tb hard drive doesn't count as an archive anyway, your point?>worst case scenario you lose it all and have to get the files from somewhere else againif it's readily available stuff, then why are you hoarding it anyways? and why do you need it to be lossless if sourcing it again is so simple?>off site backup is stupid because it isn't important dataagain, then why save it in the first place?>shan't the difference is palpable lmaothis part's gotta be bait rightlike the rest of the stuff you wrote... that's gotta be satire
>>103214539More modern codecs produce something like an "idealized" version of the input that might be considered sounding even "better" than the original by some, but it is exactly the lack of imperfections that were present in the original and were better preserved with "dumber" codec that in longer listening makes the experience jarring and artificial. Like the sound was some auto-vectorized version of something that was originally a photo.
>>103214708post X/X/Y test results or GTFO
uMurmur on my test server allows up to 128K. Try out different speeds from multiple computers. The phone client is Mumla.server: boards.nochan.netdefault port.
>>103214708Delusional. MP3 fails spectacularly on some samples even at 320kbps and produces artifacts that aren't present in the original or with high bitrate Opus.
>>103214811Changed max bitrate to 420K for uMumurClients are Mumble and Mumla. Uncensored so dont connect if delicate.
Does Discwerd use Opus?
>>103187491>hear>kbps
>>103217040I set up uMurmur so people can test Opus up to 420K>>103214811Though the client I think tops out around 128K but there might be one that goes higher.
>>103215643for vc yes, 64 kbps by default
>>103217239yeah just figured I would help people get some real world experience as the debates were going and going like the energizer bunny.
>>103199818hearing eye sight etc etc it all goes to shit eventually
>>103218309It looks a little exaggerated here.
>>103187451Personally I use 96kbps for music and 48kbps for podcasts
i've been doing 320 because i didnt know any better, is 192 really transparent for music?
>>103222183128 opus is, and you probably can't distinguish it from 96 >>103214747
>>103222183Food contaminated with mercury is transparent, does not mean its not harming you.
>>103222183192 is overkill.
