(Actual) Best Settings for High Quality Audio Playback on YouTube [Debunked!]

[@joesalyers]

A someone who helps the open source audio community with testing of the KZbin CODEC I can tell you everything you need to know or get you in contact with the actual developers. Here is my advise anyone needs for KZbin audio. Keep your audio LOSSLESS as a WAV or AIF until it is uploaded and compressed by KZbin to Opus and it will be better than a 320 kps MP3 or AAC file since a 256 kbps Ogg opus is a far superior lossy audio codec to the 320 AAC or MP3. KZbin uses VP9 for channels of over 1000 subs or higher than 1440p content. VP9 (sucks as a video compression but the ogg Opus audio codec is the best lossy audio codec period, nothing comes close.) But uploading AAC or MP3s and converting them to the OGG OPUS causes a second layer of artifacting that is difference form the MP3 & AAC protocol. OOG OPUS is open source and the reason KZbin didn't let you talk to the engineering team is because it's all community based and youtube makes donations to the teams like Mozilla and the Xiph organization who actually keep the codec up. It makes it a tax write off. Here are the names of the developers. Jean-Marc Valin, Timothy B. Terriberry, & Koen Vos. I've done so many tests with this and here is what you need to know Opus conversion is best done at -8 to -9 because the artifacting you are hearing is mostly from a level too hot and KZbin and artifacting will occur at high levels if the Opus "gain compensation algorithm kicks in after about 2000 views and tells the severs to RE PROCESS your audio at a lower level. When will you notice a difference. Before 100 views the video will be at full volume, unless you set it to premier the following day, that is when it kicks in. Your video right now has had the audio re-encoded at least twice since you were 8db too loud on your upload. Bu tif you want more info contact me and I can help if need be. Google uses the open source community to build their infrastructure, from ChromeOS to KZbin's codec and so on! KZbin-dl is a terminal application that will pull the original WebM VP9 file down from youtube, it has no GUI for Mac or Linux but I think it has one for Windows. I don't use windows so I wouldn't know.

[@RaytownProductions]

Wow thank you so so much! This is the most helpful comment I've ever received. I'll be in contact with you sometime soon. Cheers!

[@khayydilemmaz]

So should I export my audio from the DAW at 44.8/24 bit? Or 16 bit wave?

[@khayydilemmaz]

@@RaytownProductions So should I export my audio from the DAW at 44.8/24 bit? Or 16 bit wave?

[@shipsahoy1793]

@@khayydilemmaz My guess is that you do 16bit, 44.1 kHz lossless .wav file for CD-quality audio, or if you’re hell bent on high res, you’re hell-bent on 24-bit, I’d probably use the 192 kHz rate, provided any other equipment you will use with the files supports that. 👨🏻

[@MrBazsi888]

could please explain in a brief (and simpler) what I have to do in the real world to gain highest quality audio in KZbin ?

[@skumberg7160]

I did some tests for about 2 years ago. There was some differences in audio quality depending on the resolution you was watching at (when you chose different playback options), i have not tested 1080 vs 4K, but there was differences between 1080 vs 720 and the other lower resolutions. And I found out that there was differences in audio quality depending on the numbers of views for the video. The quality steps: 0 -9 views 10 -19 views 20 - 29 views 30 - 39 views 40 - 49 views 50+ views And in "statistic for nerds" can you see if KZbin have normalized the volume for the video.

[@davidasher22]

Just to update you.. I’ve done a bit of research concerning this and there is a difference between testing a video you upload privately and public video that’s been aloud to be subjected to KZbin codec. YT stated in an article this July that they don’t stream video above 256kbps. That’s a lossy codec that cuts off at 16khz. So any real world test is going to playback at this bitrate. I’ve tested 4K videos and 8k videos. Everything on the consumer end, ends up filtered by the codec. Even KZbin Music..

[@ProductionAdvice]

Interesting stuff. How did you create your mp4 files ? Because afaik they will always have AAC audio, whereas MOV files can contain lossless audio. They are then transcoded by YT to AAC for streaming & playback. So I think what you're seeing might be the difference between the encode quality of the KZbin site (from lossless audio in MOV files) and the software you used to encode the mp4 files. Or did you test different encoders ? Finally a very minor point - the spectrogram lines parallel to the original curve are harmonics, not aliasing. I think you know that, you just mis-spoke :-) There is a tiny amount of aliasing though, right at the top where the curves start to reflect back down - even in the 44.1 MOV version, which is odd. But almost certainly inaudible. Are you saying you were hearing aliasing in your uploads, though? The vertical spikes are odd - I'm assuming encoding artefacts.

[@RaytownProductions]

Thanks for the great comment and clarification to the artifacts I was hearing Ian. It's all quite odd which is why I posted this video. Something that is still a mystery is why I don't see a 16khz cut off like others report... I think you are exactly right about the origin of the artifacts. The mov was encoded lossless and MP4 was encoded with the highest quality lossy codec I could use in my Divinci Resolve Studio video editing software (192 KBPS AAC). After reviewing the video, I wish I would have done a better job explaining that more clearly. At this point from the discussions I've had with others about this video, I don't think the issue is necessarily a difference in container format (MP4 vs mov) but rather the second generational losses seen from submitting a lossy codec to YT. What amazes me is that YT recommends those lower quality setting in some of their documentation (maybe for bandwidth reasons?). I haven't had a chance to test other encoders but plan to do so in the future and will publish any updates I find. But the simplest method to high quality audio appears to be just use the mov container with lossless audio (wav). This might seem obvious now, but YT muddied up the waters with their support documentation. And anytime we blindly follow recommendations without testing, that's when we can make some bad decisions 🙂. Cheers!

[@ProductionAdvice]

@@RaytownProductions Sure, I agree - in fact I'd always assumed that if you submitted mp4, KZbin *wouldn't* re-encode it, so that's pretty disappointing. So two follow-up questions: 1 - What does the spectrogram of the mp4 you submitted look like ? If it's clean, we know double- encoding is the problem 2 - Would the same thing happen if you encoded to 128 kbps AAC ? That's the data-rate of most of the files I've seen, so maybe that would avoid a double encode (if that's what's happening)

[@connor3086]

Super interesting breakdown! Definitely need to go down the rabbit hole of research now

[@publicstar_bsk]

24mins??? lol. I don't have time dude

[@texanhoneybadger]

Thank you for this video! I have very good equipment and it was driving me crazy that the audio sounds so crappy on KZbin when it sounds great on my original file! I will be uploading my videos in .mov format going forward!

[@fredbloggs6080]

For quite a while KZbin has treated video and audio separately, which is why there is no difference between 720p, 1080p and 4K. Soon after I started uploading videos to KZbin about 10 years ago, I came upon a formula that had the best audio, and it turns out to be verified by your tests. Whatever the source of the audio, be 44.1 kHz from video from another source, or streaming, or 48 kHz audio from over-the-air TV broadcast, or my own recording at 24 bit, bit 96 kHz, I edit and normalize it in 32 bit float and export it in wav, originally at 1411 Kbps, 16 bit, and more recently at 24 bit, 2117 kbps or 2304 kbps, depending on if it's 44.1 kHz or 48 kHz. I join it with the video in MOV. Although it will never sound as good as the original on KZbin, I don't get any artifacts. By the way, I don't believe it's the MOV container that is preventing artifacts with 48 kHz, it is the wav audio, assuming it's at "lossless" at CD quality or better. The mov container is simply needed to combine it with MP4 , mpg, ts, or other video. And although I normalize my audio to 100%, I do not compress it at all and as far as I can tell it is never reencoded to a different volume level on KZbin, even after maybe hundreds of thousands or more views (different channel), where stats for nerds shows that it is normalized 100%, and the loudness is about 72% or 73% on average, with maybe an average of 2.5 dB (headroom, I suppose). Speaking of headroom that's why the high bit rate wav uploads don't have artifacts, because of the headroom the KZbin processing has to reencode.

[@Blueyzachary]

Heads up, container format shouldn’t matter, cuz video goes to mezzanine, and audio goes to PCM. When streaming, audio and video are separated on KZbin. Codec matters, so upload as PCM to avoid generation loss of any kind. You can get the streamed files with YT-DLP. You’ll have control over any encoding that the software does, as you tell it what to do in FFMPEG. You can get the Opus-encoded audio that KZbin has encoded from you, and that SHOULD be your highest quality audio for the upload. Like I said, the resolution you are streaming the video at doesn’t necessarily determine what type of audio you are receiving; only a few of the streamed formats are audio and video multiplexed, and those usually only happen at lower bitrates for compatibility reasons.

[@billystar1306]

Thank you so much. With similar tests I did, I ended up with very similar results. The best video format for audio quality is definitely .mov with audio settings (for DaVinci) Linear PCM file at 24bits. Sampling rate 44100. Even back in my studio days I thought that 48 and 96 were more difficult to handle, with more bassy frequencies here and there and worth the fuss only if experienced on hi-end systems.

[@Christian_svedin]

Great job! I know there have been videos on how to manipulate YT to get them to stream to their better quality codec. Maybe that will affect audio as well? The basic line of that was that you should upscale your videos to 2560x1440 or 4k. On a side note if you want to reach an audience beyond us sound nerds 😁 you might want to do a dumb down version and one that is more advance. Just so you get to the point faster. Great job though 👍

[@FinalBaton]

Thank you so much sir! (and Music Factory Studios as well). I will export my song/album videos in the MOV container, and will pump in losless WAV audio at the best quality I have with my masters (which are 48kHz 24 bit).

[@Caayaaa]

Can you do a data summary please? Awesome work!

[@Jim951cup]

Hey Bobby what's up? Just wanted to clarify something about the null test: when you get phase cancellation it does prove that both files are identical. However, when you hear a difference, the sound of the difference is not the same as the difference in the sound. It's easy to verify with meaningful content like music for instance. Compare two different revisions of the same master and you'll see that the difference you hear in the null test doesn't necessarily match the changes you actually made during the revision. Otherwise, great video: you might be onto something really important here.

[@maxmazmaz]

Hi Jean, anytime you mix down a song that has time-based effects such as reverb and delay, a second mixdown of the same song with no changes will not null out either. It's the time-based effects that cause that.

[@dadw7og116]

When I right click your video and select "stats for nerds" I see opus (251) for the codec. Did that change when you uploaded your different samples? I mean, did they all end up being served back with the same codec? FYI, "251" (supposedly) corresponds to 128kbps 48KHz Opus audio in a webm container. "140" refers to 128kbps 44KHz.

[@_quikmelter00_72]

I freaking love you man

[@KeepTheGates]

Are you sure it's not because the shape knob isn't at 96?

[@RaytownProductions]

Hahahahaha! Nailed it 😉

[@RaytownProductions]

Only a true fan would know this reference 👨‍🔬🤘🤣

[@Shred_Rocket]

Perfect analysis! Thanks, Bobby great work as usual.

[@RaytownProductions]

Thanks so much! Glad you got something out of it. Cheers!

[@Rightchickenwing]

I'm trying to find out the best way to get the best audio quality from songs so I can then record them as backing tracks for my guitar covers. I've heard that adjusting the video from 480/1080p etc does not matter for the sound quality

[@laurant4282]

Commenting for the sake of the algorithm, because I think this is important!

[@rickbearup9514]

Great job Bobby. This should go viral! Now I wonder if there are any differences in an MOV file created on a windows machine vs a mac. They should be exactly the same right? Hmmm.

[@RaytownProductions]

My gut instinct tells me it should be the same since mov is a container for pcm wav files which should be universal whether Mac or PC (wav files made on a Mac will sound the same on a PC). But I haven't done the test yet! Thanks for the kind words by the way 🙂 it can only go viral if it's shared haha 😉

[@tutupachamama]

Anyone able to answer simple question? Whats the best audio format to mix in a video and export it Highquality to KZbin?? Thanks everyone

[@yeshuacreates]

Hi @RaytownProductions, one important thing that seems very unclear in your discussion of MP4 vs MOV is what codec you're using for those containers. For example they could both be used for h264 or h265, and MOV could be used for those plus others like ProRes etc ... Then I didn't notice you talk about bitrate choices either... If you have any notes about that, I'm curious. This plus frame rate choices can have a big impact on the relationship between audio and video with music videos or concert footage...

[@tarno_bejo_]

Is there a new update from this vid? Feb 2024 I ve tried this kind of thing on my other channel. I think, it depends on the coder of the audio part of those videos. If you re using aac, neither mp4 or mov will give you the best result. Aac also depends on the bit rate. 48kbps will filter the audio around 11kHz. 192kbps will filter the audio at 16kHz. 320kbps will filter the audio at 22kHz. Higher than this sample rate is the same. Even with 512kbps, this aac will filter your audio at 22kHz. Those filters are the ones caused artifact. Like, im not sure about the video editor you were using. Even if you have mov file but it has been coded with aac (using videopad in my case), it will filter the audio as bad. But, the mov file can give you option to code the audio with PCM uncompressed as i understand (like using adobe premiere pro 2022). I mean, all of those happened even before uploading (the raw files). KZbin converted the audio part into vorbis as i understand? Then as today, it would filter out anything above 20kHz. Since most music are under 20kHz, artifact should be minimum or even non existence.

[@samueljholly]

been struggling with this problem for a while. spending countless hours on my craft just to have it saturated & watered down by youtube floors me. i’m confident in my masters…always. but i have to be doing something wrong in the process of uploading to youtube because the youtube upload is always quieter & just doesn’t hit the same, especially compared to the original file, but also even compared to other artists’ uploads. it’s a hurdle i have to clear

[@NilakshaFreeson]

This is by far the best audio engineering in depth illustration I've ever come across! So logical! U r more of a scientist than an audio engineer! Cheers mate!

[@RaytownProductions]

This might actually be true... What if I told you I AM a scientist?!?😉

[@NilakshaFreeson]

@@RaytownProductions Awesome! Same here btw! ;)

[@RaytownProductions]

Not to get off topic, but what type of research?!? I'm a chemist 🧪👨‍🔬

[@NilakshaFreeson]

@@RaytownProductions Wow. Awesome. I'm a Molecular Geneticist and a Bioinformatician ✨

[@Sciencegrinds]

I do not like the sound on this video. I notice quite a few podcasters use similar sound settings to produce this timbre of voice sounds. I am a trained classical musician but I am an older person. I tended to have a V-shape on the 1980's manual equalizer settings of my HiFi system, whereas my untrained contemporaries would have an INVERTED V-shape on theirs, so my taste in audio settings was different from my contemporaries. This applies to sound from radio, vinyl records and eventually CDs. Now with digital devices I find some settings, including those used in this KZbin video, very uncomfortable to listen to. I don't know why? Or if there is anything I can do on my phone or TV to change these sounds like I used to be able to do with the equalizer levers on my stereo system? Guidance would be much appreciated.

[@ClayWheeler]

@Raytown Productions For the record, between 2020-2021 KZbin had to temporarily Transcode all videos being uploaded in this time period in Slightly Lower quality (both Bit-rate and Sample-rate) Because you know why? COVID. Due to the Pandemic, KZbin server overwhelmed by insane jump of Video uploads all around the world. So the transcoding server need to be adjusted a little bit to compensate this trend. Luckily, today, not anymore. Also, the Pinned comment is 100% Correct.

[@sonario6489]

Meanwhile I'm sitting here wondering what sample rate and bit depth I should use for audio for animations, though the audio is gonna end up being added in post-production because audio in Source Filmmaker does noooot work on my laptop.

[@RaytownProductions]

I would suggest 48 khz and 24 bit for that application. Let your post production team dither it to 16 bit if need be. Better to have these settings to "future proof" your work as standard for video is 48 khz. Keep the audio as pcm or uncompressed wav files format and not aac or mp3. Hope that helps!

[@arturts987]

Thanks for the content! I'm wondering, is there a posibility to improve audio quality when live-streaming to KZbin? The only codec that seems to be available is AAC-320 kbps.

[@SpacedOutDoonie]

Damn bro this is like the best detailed video about this subject I ever seen, good shit.

[@MrHeHim]

I've noticed things at Google aren't doing so well in the last few years in some departments. Feels like committee decisions where nobody understands what's going on. Held together by tape and if it ant broke attitudes when it's not working right in the first place. Everything from menu and button functions on my Pixel 6 to strange/frustrating changes in Google search when shopping. Nobody even wants to admit there's anything wrong to avoid being blamed altogether

[@sleepsoundly4728]

Hi, when you say that you used "Pcm files" are you refering to Linear pcm right? I read that Linear pcm starts at 48khz instead of 44.1khz, but it makes me wonder if i have a song at 44.1khz and i make a video using that song and exported as Linear pcm, with it create artifacts because it's 44.1 instead of 48khz?

[@davidasher22]

Did you notice a brick wall filter at 16k? It doesn’t look like it on your graphs. But I recently tested some KZbin videos after hearing from a tech KZbinr that anything above 16k is filtered out and I verified it with some of my own tests. I ripped the audio from 3 different music videos streamed at 1080 vs the same songs ripped from iTunes. The KZbin audio was missing everything from 16k and above. If you’re not getting the same results then what do think I was seeing? Just for the record, I connected my iPad to my MacBook via iDAM to rip the audio.

[@IsaiahPrather]

That’s moreso a sign of MP3 quality, than what youtube is doing to the file. 😄❤️

[@davidasher22]

@@IsaiahPrather the samples were all taken from official videos from professional artists at the highest quality playback. I only did the test after I heard from a reliable source that KZbin filters all their audio at 16k. And like I said, I did the tests and saw it for myself. I have a theory that it may have something to do with loudness penalty procedures. It can take some time before YT engages its loudness regulations so maybe that’s also when the filter the audio. That would explain why his test tones are unaffected and a Billie Eilish video is. But this is just a guess. Maybe I need to test more than 3 videos. Idk.

[@RaytownProductions]

This is a really good point. I double checked to make sure, but NO, there was NO filtering above 16 kHz for me. 20 - 20 kHz was clearly all still there in ALL the test files. I apologize that some of the spectrographs don't have the scales present, but a few of them do and you can clearly see they all go to 20 kHz. So either the algo has changed OR the tech YT person was using software that transcodes the audio without them knowing (which is EXACTLY why I captured the audio the way I did). A way to double check would be to export just the audio from your YT videos and look at it with sonic visualizer to verify that high frequency information is there before you upload to YT. It's also very possible that if you downgrade the streaming quality to 720 that the audio degrades - I didn't check that... Great question! Thanks for watching. Cheers!

[@davidasher22]

@@RaytownProductions when I did my test I did check with a sonic visualiser. I used 3 different music videos from Billie Eilish, Foo Fighters and MGK. I do remember hearing that YT doesn’t immediately adjust for sound penalty levels. That it takes like 1000 views before they turn it down. If that’s true than possibly the filtering happens then too. That would explain why your test video were unaffected. I did my tests last month so I don’t think the algorithm changed. When I saw your video I was puzzled because I clearly saw the sine sweep reaching 20khz. I guess I dig a little deeper. Maybe I need to compare a recent upload to something with views and see if the sound penalty thing is true and if that also determines the filter. I’ll let you know what I find.

[@firmans12]

KZbin sounded much better now for the video quality. I remember in 2010 ish when my internet is slow as hell before youtube introduce 144p. 240p will drop the quality of the audio just like mp3 with 128 kbps or below lol and 360p sound much better. But now luckily youtube don't make 240p sound bad anymore lol

[@onemic-theminimalist]

What is the noise floor of the spectrogram? Is the noise at -150dB or -20dB?.. Without that half of the info, alias numbers are useless. We need to know if the aliasing is in the audible range of hearing.

[@Reticuli]

Be nice if the Stats for Nerds showed the audio khz so you could manually change your OS/soundcard to that sample rate.

[@SGC-]

sounds like they are doing the same thing instagram is doing, they use an mp4 codec that uses the lowsest settings possible and converts any stereo sound to mono and crushes it on top of that. i used to be able to upload there just fine but in the last year its gone to absolute crap, i basically have to ruin the track for it to sound decent on instagram

[@RaytownProductions]

What settings do you use in your KZbin videos? And how does the audio sound?!?

[@heavymetalmixer91]

Offtopic: I don't remember if you mentioned this, but I dind't see you testing the 16 bits files. Would the results be the same?

[@RaytownProductions]

@@heavymetalmixer91 I believe so. The differences would just be in the noise floor aka dynamic range. They might have some dither added, but it wouldn't be significant at all. Thanks for the good question man!

[@OzanYarman]

You are absolutely amazing. Thank you a million bunch! I now know that Mp4 and M4v video formats are butchered by youtube and that is why I am having the worst experience with 48 KHz sampling rates! I suspect the reason behind this is to have people prefer mp4 and m4v so that the cheapskate youtube can conserve processing speed when transcoding these formats. This is a huge scam!

[@fb510m]

i'm only watching to try to figure out how to LISTEN to videos in the highest quality audio

[@john_mclucas]

mad helpful rn as I'm focusing on handing off more YT stuff!!

[@RaytownProductions]

So glad you found it helpful! More to come on this topic on other platforms! Stay tuned....

[@DeborahLevoy]

So well done and interesting, thank you! I realize I'm coming in late here, but I have a question I'm hoping you might answer. I am using my friend's GoPro 9 which only seems to shoot in mp4. Does converting the mp4 to mov after shooting help at all or is the point to shoot in mov in the first place?

[@petrparizek9945]

I'm sorry to have to inform you that I've just managed to make a sine sweep sampled at 48 kHz, encode it in AAC, put this data into an MP4 file, upload the file to KZbin, and get ***NONE*** of the audible aliasing that you were getting in your examples. Are you sure that some of your intermediate files, during transcoding or resampling, are indeed free of audible aliasing before you upload them to KZbin? If you wish, I can send you a link to the actual MP4 file, in case you want to test for yourself.

[@RaytownProductions]

That's a good sign! Maybe KZbin fixed the issues on their back end. I'll have to try to reupload my test tones and see if the artifacts are gone. I know there is continuous development with their streaming platform. Thanks for the comment!

[@WinonaCellist]

Great field test. Thank you for sharing this.

[@firmans12]

Love youtube audio now more than spotify for the same song. idk if It's a different track uploaded but i tried the official youtube music vs spotify. in youtube the transient is just feels better and more open and punchy while in spotify the transient is tamed too much IMO

[@richh650]

Excellent video sir!!! Thank you for sharing it.

[@takeshi_shinyama]

Appreciate it so much sir🙏 ive been struggling with this for a long time. Really really appreciate your work and sharing with us! One question, what about the difference between 24bit and 32bit depth sir?

[@RaytownProductions]

I don't think that will matter. 24 but should be more than enough 🙂 cheers!

[@takeshi_shinyama]

@@RaytownProductions got it, appreciate you sir. God bless😌🙏

[@shintsu01]

i wonder if someone can help me on the playback issue i have with you tube, it sounds way to often as if the mic is clipping for the videos i am playing, since its happening for so many videos i feel it has to do with my audio system. however for twitch it works fine and i dont have this issue so i assume it has to do something with youtube or my browser combo

[@piyasirimusicproduction]

I think that .MOV sounds way better therefore I dont use .MP4 on my channel. Thanks for the confirmation! 😁

[@शिव_सागर]

Thank You Very Much ! 🙂🙏

[@MasterDuelMax]

Its really good to see Johnny Sins get out of that toxic industry.

[@metalvideos1961]

so i am just wondering. if i uplaod an album to youtube. is there a difference if i upload lets say a 128Kbps MP3 file (convert it to MP4 first) or a 320Kbps file (also convert it to MP4 first) both in 1080p. is there a difference in audio on youtube?

[@RaytownProductions]

I would think so because you are getting second generational losses from YT when they re-encode the video. Safest bet is to upload the highest quality audio possible, which would be a lossless wav or flac. Mov containers can have lossless, whereas MP4 often is difficult to get the lossless file in that container. You usually have to mux it yourself with a 3rd party program (Divinci Resolve doesn't support it and I also don't think premiere does either). Hope that helps!

[@metalvideos1961]

@@RaytownProductions yeah well i use the simple version of windows live movie maker. nothing special. he can however support WAV files. But i have tested it. and no matter if i use 128Kbps mp3 all the way up to WAV with the max bitrate. the output file still is the same bittrate. like around 190Kbps MP4 files. anyway thanks for your comment. i see that it doesnt really matter THAT much.

[@RaytownProductions]

@@metalvideos1961 nothing to lose sleep over, that's for sure. Cheers man!

[@luiggimondoli]

Awesome work! Thank you very much.

[@GordoEats98]

Love the vid mayne

[@hdmoviesource]

Great video and thank you.

[@cozydaybit]

Really appreciate this

[@AndreasMSchroeder]

Liked commented and subscribed. KZbin need to listen to you!

[@RaytownProductions]

I tried lol! Well hopefully this is a step in the right direction 😀 thanks for the comment!

[@MrRafalSC]

A very interesting experience !!!

[@yourforeignlocal]

17:36 Will it be the opposite if those sound files were captured in 48kHz?

[@RaytownProductions]

Nope, regardless of how I captured it (44 vs 48 vs 96 kHz) the artifacts were all the same. Great question!

[@Reticuli]

@@RaytownProductions What rate was your OS set to for KZbin, and is KZbin converting all audio frequency sample rates to one common rate like 44.1khz?

[@rocketman374]

The null test between 4k and HD just sounds like dither noise doesn't it?

[@rocketman374]

Also, aliasing is that little trace that goes up and to the left right around 16k. The rest isn't aliasing, it's random garbage from lossy encoding. Did you encode your MP4 with AAC at 48? If so, you encoded it once, then KZbin not only did it again, but SRC'd it to 44.1, which is their outbound bit rate for AAC. Did you catch the AES video that Bob and Anna did about streaming. KZbin had a rep there explaining the mess that is KZbin. Lol.

[@rocketman374]

But as to why 48 is much messier than 96 in MP4, I have no idea. Lol. Did you ever figure out why?

[@ellisdidasong]

So what are the "(Actual) Best Settings for High Quality Audio Playback on KZbin"? Even a summary of your recommendation, "Do the experiments yourself" is not an answer

[@ClayWheeler]

it is depending to the Uploader themselves including "You" Basically if you uploaded a video with cheap audio, it will sound trash. That's it. Just upload your video but with Uncompressed Audio. Like WAV or AIF. See pinned comment

[@tikaal]

what about codecs?

[@RaytownProductions]

Good question! I should have mentioned this in the video. Everything followed YT upload guidelines. So all videos were H.264. I didn't try all the different codecs because there are so many. But if you have a good experience with one over another, let me know and maybe I can check! Hope you found the video helpful!

[@BarryMaskell]

Most likely they wish to push their music platform

[@cemski]

Sorry, at two minutes you still talk around what you are about to tell. My patience has changed in the last years. I tend to save my concentration for direct infos. I am sure you have good points and knowledge!

[@ClayWheeler]

Conclusion: The Video you upload MUST HAVE: Video Codec: Any, like H264, H265, AV1. 8 Bit or 10 Bit doesn't matter Audio Codec: WAV, PCM, OGG Vorbis, all at 44.1 KHz or 96 KHz 24 Bits. AVOID AAC & AC3 and 48 KHz, must NOT Louder than -8db Video Container format: MOV . Avoid MP4 MKV I know this comment might be outdated soon, but at this time, those are the best settings for KZbin video upload

[@insertanynameyouwant5311]

ouch I always did 48kHz and MP4, any advice for software that exports MOV? Handbrake only does MP4 and MKV

[@RaytownProductions]

Mkv should be able to hold 24 bit pcm wav files! I didn't test it but I bet that would sound equivalent to mov. Hope that helps!

[@insertanynameyouwant5311]

@@RaytownProductions oh yes, thanks, if MKV will do, then it`s awesome!

[@RaytownProductions]

@@insertanynameyouwant5311 just be sure that when you export your video that it's set to PCM Wav file and I think you will be all set. Cheers!

[@insertanynameyouwant5311]

@@RaytownProductions sorry didn`t get notification, I`m afraid I don`t see that option. It offers various formats, but not PCM. Do you think FLAC 24 bit would work?

[@vortexalerand9949]

RE broad mp4 recommendation, I'm thinking like this; one could assume that a general majority (?) of YT's uploaders don't know or understand the option or care greatly, and so, for KZbin with storage and servers in mind , it could makes sense to set quality at a rounded minimum , while also giving you the choice to go up from there, maximize and therefore raise process requirements, theoretically I base this on whether on my premise of server/storage load is actually relevant , has something to do with it or , or maybe did once upon a time, or a complete nothing sandwich So I genuinely don't know, what are your thoughts on this?

[@Exert145]

22:49 I see very small difference in 96khz compare to 44khz and 48khz sampling rate, just look closely.

[@Reticuli]

Which means it's very likely KZbin is either converting everything to a common frequency sampling rate and/or his OS was on a single rate and he didn't change his OS rate to match the video.

[@scottwaddell298]

Would be cooler if you just got to the point in a 4 minute video...

[@davidorlins7249]

If you want high quality audio, KZbin is not it. Get Tidal or Prime Music. I'd never listen to music on KZbin.

[@pugazhendi1514]

Talking too much