I've had people asking me for FLAC files of a project I recently did, and I didn't really know why. I've had no problems with MP3's for many years. So I exported a full FLAC file, like no compression, the file is 200mb, it's insane, and A/B tested it against the normal 8mb 192kbps MP3 file in Studio One. I could not tell a difference on my monitoring headphones, whether knowing which was which or not. I'll still do the FLAC files for other people if they really want it, but compressed files sound perfectly fine, to an extent. I know how terrible low quality MP3's are, I used to have many of them, and some of them are still kicking around in my library since they're very rare songs. Bumping up to 320kbps doesn't really change the file size for MP3, so I've moved up to that, but I'm still going to listen to MP3's and be content.
Thank you for the video. Because the encoders are very good these days, it is generally really difficult to hear, especially in higher bitrates. But keep in mind, the audio codec here on RUclips is already lossy, so we can't never hear the real uncompressed WAV audio in this video.
LAME 3.99 & newer are pretty much reach transparency at V2(192kbps) I can't tell that from Lossless. But I do encode some edge cases at V1 & V120(Helix MP3).
I did these blind tests a few years ago and I honestly couldn't tell the difference between 320 kbps and the wav file. and when I was just choosing based on what I thought sounded better I was predominantly choosing the MP3. That was very humbling and nudged me down a rabbit hole of digital audio and audio science in general. and its been a great ride! if only more engineers and audiophiles would try these test themselves. its only made me smarter, not to mention has save me some money! great stuff!
@@DicksonDaggerCould be a problem with your DAW's export feature. If it doesn't use LAME, uses an older, outdated version, and/or is double encoding (re-encoding multiple times) it will degrade sound quality pretty quickly. I always export from my DAW at lossless and/or with lossless stems, just to be safe. And then I encode later with a tool like Foobar or fre:ac. This maximizes quality, keeps a lossless version of your music for archival purposes, and makes it easier to share the mp3 with others!
@@TheDylandProductions My point is your track is going to sound better because you started with lossless files to begin with. But when you start processing lossy files like mp3, that file gets recompressed again if you export as mp3 and you risk losing quality. I’ve had the my mixes sound great in my DAW using mp3 beats but once I export it as mp3 the signal distorts in a weird but subtle way.
RUclips down samples audio to at best.... 160kbs (Opus 251). That is what this video's highest audio quality is at its maximum. YT is veryy sad when it comes to throwing away the large majority of all the original data the video maker wanted.
I asked my producer friend (he produces electronic music, and I mostly listen to this kind of music so i know what it's supposed to sound like) to export the same track directly from the DAW in a couple different formats (@ 44100 Hz, since higher sample rate =/= higher quality), and we listened to them in parallel firstly blind, and afterwards with knowing which are which, and here are our impressions: Lossless formats (WAV, FLAC & AIFF, 24-bit and 16-bit each) - We could pick out the lossless formats in the blind test, and we agreed on linking the sound quality of these files the most, but we couldn't tell them apart from one another, even after multiple listens... 16-bit FLAC-s have the smallest file size, and plenty of tagging capabilities, so this would be my pick for Hi-Fi listening, while sound engineers, producers and DJ-s should go for WAV for compatibility and editing reasons. Lossy formats (MP3 & AAC) - We could tell the lower quality of these files, but here I could also place the lossy files into 3 tiers of quality between themselves, while my friend just said "these all sound terrible", and stopped listening 😅. -1st tier is 320kbps MP3-s with CBR and "Highest quality" MP3-s with VBR: I couldn't tell the difference with these 2, but the latter has slightly smaller file sizes. These 2 are perfectly fine for entry level Hi-Fi (eg. $20-$50 IEM-s or speakers), and is what I use for my library since the file size diffrences between this and lossless are about 60%, with the quality staying really close. -2nd tier are 320kbps AAC & 192kbps CBR: Louder parts sound identical to the first tier, but quieter parts of songs is where you can tell there is information loss. These are a hefty 80% smaller than lossless and are also fine for average listeners, but I prefer HQ MP3-s. -3rd tier are everything else (128 CBR, 192 & 128 AAC and MQ & LQ VBR): All of these don't sound good to my ears, although the difference is harder to notice once I tried some lower quality earbuds. I'd stay away from these, unless you're desperate for storage space. While this tier list might suggest otherwise, the diffrences between all of these weren't THAT great, and to someone who didn't take this kind of critical approach to listening, they might have been indistinguishable (except the 3rd tier, where I think everybody would be able to tell). Also keep in mind that this test is done on electronic music where audio is pushed to it's limits compared to other genres, and I imagine the diffrences would be even harder to tell for pop or classical music. PS English isn't my first language so forgive if I have some spelling or grammar errors.
The digital equivalent of analogue sources 30+ years ago : good quality = cassette / excellent quality = Vinyl LP / Studio quality = Reel tape (at least 15ips), but most consumers then listened to cassette through headphones so enjoyed the worst version of the master, so the majority of music consumers taking streaming now via spotify etc is getting a pretty decent sound if you compare it to the mass consumption of the cassette back then.
Initially, in the early cassette releases, but as they became more main stream, the quality massively improved. I'd prefer a high quality cassette release to mp3 any day. The only downside to a high quality cassette was the tape hiss, but then mp3 carries with it, it's own inherent noise. As for vinyl, it used to be the best, but as with cassette, its sound was degraded overtime. It increasingly crackled. The best sound is still CD when the audio is AAD, rather than DDD as you then get the best of both worlds. Providing that is the record lable isn't lazy and simply re-releases the original vinyl master.
Cassette is better than flac and MP3 because it has ultrasonic frequencies. MP3 and flac have nearly all the ultrasonic frequencies stripped out of them. It's why vinyl is still popular.
Well done. I've always been able to hear the artifacts, and this demonstration really brings it home. I agree that most people listening to streaming in their car, or on cheap audio gear will likely never really notice, especially without the A/B comparison. When working on a mix, or listening on my high end stereo equipment, the MPS really bugs me. I have a client who sent me his tracks recorded with MP3, and I just could not make them sound right. I made him render to WAV. After he did, I had him A/B and he could definitely tell. Lastly, i heard some pumping in the left channel cymbals. It's there is the WAV, but it got exaggerated when compressed. Anyway, great video. Thanks!
Great video. Very interesting. I have done some listening tests between AAC and Apples's new Lossless formats. Only the Max 24 bit 48 KHz lossless stream through the Apple donegal D/A to my Sennheiser and RODE headphones. I can't tell the difference even though I can still hear up to 12KHz (I could hear up to 17 KHz in my 20's).The AAC compression noise (on your video demonstration) sounds higher in frequency than the MP3 noise. Defiantly better I think especially for my 70 year old ears.
There is an ABX plugin for Foobar2000. You can encode files at different rates, then test yourself. In Archimagio's 320K MP3 test people could not tell the difference.
The AAC codec should work better than MP3. Both codecs were invented by the Fraunhofer Institute. MP3 came first and AAC was created to overcome deficiencies in the MP3 codec. Given your examples, I believe they succeeded.
I dont think it is adding noise, the noise is the sound of the difference. you can still kinda hear the vocal in there even on the high quality. I think it is total cancelation, like a null test, except where lossy format has removed some information.
I like listening to early 2000s music like Staind in mp3 because that’s how I used to listen to it when I was growing up. Either that or Flac (CD quality.) but for 2010s music I use RUclips and use the original videos that I listened to lol. Because that’s why I did in highschool. The artifacts add to the nostalgia for sure. Its like tape distortion on an old record or crackle. When I listen to 30s music I use vinyl rips. Always use wired headphones or earbuds. Cuz that’s what I used to do. Lol maybe I’m crazy but it helps with my anxiety and producers even use this technique in their samples to create certain a certain timbre. I think people need to realize Bluetooth is ruining their music way more than these lossy compressions because it is also lossy compression but in real time. Which is much harder and causes audible artifacts that don’t sound good really. Unless Ofcourse you grew up listening to that 😂
I used to collect or create my cd albums as FLAC files, but now I have a free disk space thanks to you 😉. Since listening to all of those different LOSSY formats didn't make so much difference why keep FLAC files, right? Cheers
The difference is VERY subtle, the aac sounded a tiny bit clearer vs mp3 which sounded a bit more noise-y to me. IDK if more noise makes for better quality or vice versa. Thanks for the info!
Are you sure that's the noise that's being added? I would assume what you hear is the delta and not what is added. Meaning what you hear is what is missing, not what is added. But I don't have this plugin so I don't know if it's actually what is added but seems a bit strange.
I could hear a slight difference. The punch and/or depth seem to be noticeable when using the mp3 or aac. But, I'd love to get the Fraunhofer and do maybe more comparison with different types of music as well. I think some codecs maybe better for certain styles or genres of music. With me getting into doing more mixes/mastering, I'm wondering if I can find a way to mix/master for the betterment of these codecs and etc. All in all, it may not even be worth the time to invest, as different sites offer different lossy audio anyways, it is what it is, lol. I think I was just disappointed noticing a Video mix I did to an AAC format, I could hear my own difference, and was concerned. So, it brought me around to this video, lolol! Thanks for the video!
What I noticed is that AAC LC stayed closer to WAV. it slightly brought the vocals and kicks a bit more forward but not by much. the MP3 files heavily brought forward the Vocals and Kick drums and made them sound more up front and centre, the track became slightly unbalanced and lost it's space and depth, some frequencies clouded over others.
I know there is a difference in sound quality, but I don't really hear it unless I focus very hard on it. I keep using mp3 320kbps most of the time which is a good enough quality for me, I feel like mp3's have a slightly punchier bass which I love, especially over aac which sounds to have a bit less bass (could also be just in my head lol). Personally I would only go with FLAC if I had a crazy good expensive audio setup which I and most people probably don't have. Good video, thanks.
320 is not bad but as an engineer I can hear the artifacting in 128 and its impossible to ignore. For just listening who cares? most people cant really hear it because the dont know what they are listening for. To them it's part of the song.
You have your information backwards. The program you're using is doing a null between the two audio codecs, so the noise you are hearing during the null is the audio being removed during the compression NOT added. The difference between the two tracks.
Note: ALL digital audio formats are compressed, WAV FLAC included, unless you have the band playing live in your livingroom. Modern music sounds so bad compression is the least of your worries. mp3 can go much higher than 320k, ive heard 680k mp3 and it sounds superb The differences between WAV and mp3 below 320k are that the stereo separation gets bitcrushed.
I don't think it's noise being "added" through the conversion but rather a lot of that noise is what is taken away, hence hearing the difference p.s.: the "seamonkeys sounds" being present (once listening to the mp3 itself and not in difference mode) could be the same modulating pump of that noise taken out leaving you hear more high mid-range pumping weirdly
@@RaytownProductions For what it's worth, I often had that lingering question every time I do a null test: is it something added, subtracted, or both. I wish there were a clever way to know (or maybe I just didn't think of it yet)
A lot of what you’re hearing isn’t necessarily a reduction in transients, but a changing of them. Lossy audio codecs have to quantize time information to an extent. Sometimes there are very slight “pre echo” artifacts that result from this, and will show up as the slight “transient smearing” effects that you’re describing when you’re trying to do a difference between files. The reason that this happens is technical. It has to do with MDCT windows, which are used to encode frequencies present in given time frames/“windows” that are divided out from the original file. These aren’t perfectly precise with time accuracy, the smallest windows available are 128 samples on AAC or 192 samples on MP3, which means that it will never be able to encode every transient with sample-for-sample time accuracy (even if they’re not really “losing” much information in terms of frequency content). It’s just a trade off as a result of how these lossy codecs work. There isn’t really a whole lot that can be done about it aside from using different methods to compress, but MDCT transform algorithms (MP3, AAC, Ogg Vorbis, Opus, etc) are pretty effective compared to other approaches for compression.
I feel like this track (although it's killer btw) is a poor example to use for something like this. It already has some distortion and lo-fi sounding elements.
Great video, I been uploading music mixes on RUclips and always wondered which would be the best sound recording aac wav MP3 etc etc, I'm still kinda undecided which is best wav and aac, but your explanation was great.
Yesterday, as it happens, I was listening to CSN&Y's Deja Vu album, which I had ripped to MP3 a few years ago. I, usually, listen to the CD and had not heard the rip for a long time. I heard some nasty crap, coming out of the speakers, and was pretty shocked. I re-ripped it to FLAC and it was completely cleaned up. So, was it the change in CODEC or was it just a crappy rip to MP3. Ironically, I have been in the process of re-ripping a lot of my music to FLAC, as I feel that it is, significantly, better. I, also, tried WAV but, in truth, could hear no difference and WAV takes about 5 times longer to rip, so FLACit is.
It might be my shitty earbuds but I actually kinda liked the 192kbps conversion better than the uncompressed. It kinda gives it an analog sort of warmness.
There is a fine line between objective and subjective sound quality. There are many variables, such as headphone quality, a person's age, environment, etc. Remember the Yanny vs Laurel argument from a few years ago? About half of the people heard one sound and half the other, and it suggests that everyone hears slightly differently. Back in the early 2000s when disc space was a premium, I encoded all of my music in wma format @44100hz 192kbits and now I'm stuck with it, unless I reencode everything. The last few years I have used FLAC, and that seems to be good enough for me. Interesting and fascinating video.
Not only do we hear differently but we are used to different sound signatures. We have a different idea of how an instrument is supposed to sound and how songs are supposed to sound. It even takes a little while to get used to new audio gear, which makes it hard to discern if it's different in a good way or a bad way. If the files we use have issues, like the ones cause by loudness war (low dynamic range, distortion, artifacts etc.), then that creates a bottleneck and a possible distraction. Perhaps better audio gear makes it easier to hear certain artifacts? That way the blame could be put in the wrong place. We are subjective by nature and since we haven't heard our playlists in the studio and on the gear used to produce it, our imagination can only go so far and as a result we are susceptible to biases and subjective experiences. Especially since different ears don't hear the same thing from the same audio setup. Blind audio comparison tests also suffer from these issues. If you pay attention to the volume, it can make you perceive better quality. If you listen to high hats, when the difference is in other places, then you'll be similarly distracted. So knowing what to listen for is not a bad thing, especially not when you consider how easily it is to make tests where you won't hear a difference between 128 kbps and lossless FLAC. For instance using audio files heavily compressed from the studio that sound bad to begin with and of course using low-quality audio gear. But make sure there aren't any bottlenecks in this long and complex chain, you'll definitely hear a difference even if you don't know what to listen to. Compression encoding also alters the frequency range and its algorithm can prioritize certain aspects. I find it interesting to consider the implications of the algorithm's removal of noise, grit or other sounds that are located outside of the included spectrum. Some codecs even stops at 12k. I think it's plausible that factors that are unrelated to the lossless format itself can cause false results in codec comparisons. Like I said, if you listen to the wrong details or have the wrong assumptions - you risk being misdirected into incorrect interpretations.
In all honesty listening to the track in the different codecs (on a tablet I couldn't hear a difference) I will check it out on my various monitors to see if I can hear any differences. I would like to make one point though, how many music consumers listen on ear buds or less than optimum speakers and in less than ideal environments? My guess is most. It is a sad shame that audio mixers/producers put so much effort and time into trying to create good or even great sounding music, only to have the online music streaming services slap their work through these codecs.
While AAC was made to replace MP3 in 2003 the open-source community worked on the LAME MP3 encoder all the way up until 2017. Long after the Fraunhofer Society abandoned the format for AAC. It absolutely guts the original 1998 Fraunhofer codec and AAC in all but the lowest bitrates. It's pretty much been universally adopted in every program that uses MP3 these days. Even Fraunhofer uses it over their own codec. In many ways, it's an obsolete format but it was polished to absolute perfection by dedicated fans to remain competitive.
I'm not sure that is the best type of music to judge this on. To me it just sounds like a distorted noise with no dynamic range. Try a REAL bass and drums and a piano maybe. Don't brick it, chose something with a dynamic range. The differences will show up much more.
What the codecs basically do is slice up the sound into segments and reduce the signal to noise ratio in each such that it follows the overall level per frequency band and you always get roughly X dB of signal. Whether something gets added or taken away is a matter of semantics. Add something with a negative sign. Negative birth as a euphemism for death. The codec may zero out a band completely, taking it away, but over a human perceivable time scale it's still a burst of noise. Most people listen to compressed audio without having a choice, on the radio, in video games, over wireless headphones. 128 kbit/s is really bad and the sound gets worse every time it gets reprocessed, such as when the radio station loads it, transmits it over their lines, then it gets into headphones or when people encode it for their collection. Even if it sounds acceptable after one generation, it's better to start with a better copy.
This is really really useful info...but RUclips won this battle--the data is there, we just cannot hear it. I think either high-quality headphones or being in-studio with you would be the best way to hear it. I'm an audiophile and I could barely make out a difference, however this is great to know!
The point is you don't just listen with your hears, your whole body listen and your cells are catching the frequencies you can't hear. Check bruce lipton biology of beliefs.
What you meant right after the 3 minute mark that streaming audio services use 128kbps MP3, wouldn't it be AAC? Was it a mistake? I mean no way stuff would sound so good with 128kbps MP3. Only if back in the day when I listened to 192kbps MP3 they were very bad rips, because until something like 256kbps I can definitely notice problems with MP3. Anyways, going to watch the rest of the video, seems pretty cool regardless of my question.
He's using the Fraunhofer encoder as a part of the plugin called Sonnox Fraunhofer Pro-Codec Plug-In. Don't ask me what Sonnox is, but the encoder is Fraunhofer. ;) The codec used is Fraunhofer AAC-LC with CBR (Constant Bit-Rate). Not all encoding parameters are visible. Different companies have their own AAC encoders and you don't need this program for encoding, if that's what you wonder. This plugin is just an interface for the Fraunhofer encoder.
MP3 is removing things we can't hear - by law of physics or by our ears or by your equipment. The people at Fraunhofer were scientists, they knew what they were doing. People crying about MP3 "removing things" are surely also crying about their BluRay not containing an infrared video track. You could not see it, but hey, "at least it's there, for the peace of mind". I have read extensively into MP3 and other lossy codecs and they should be complemented on how great and clever they work instead of just "zipping" the raw data. It's like removing empty pages in a notebook and only carry the written pages - I prefer that to having GBytes of data information that we BARELY can hear.
You are correct that the ultrasonics are stripped out of MP3 and even flac, but you are wrong when you say it doesn't matter. Those engineers were stupid, not clever to do that. It's why vinyl is making a comeback as it contains a lot of ultrasonics.
is the difference heard in this software taking in consideration psychoacoustic auditory masking or is it just a substraction of the 2 signals ? if it is just a substraction this is not very usefull as it will mostly expose what a human ear cannot hear :)
when i take a free online blind test of uncompressed vs compressed audio i cannot hear any difference is it my ear limitation ? my ear training ? my gear that is not resolving enough ?
Depends! I have a very hard time picking out 320 kbps from wav. In fact for me it's statistically random (e.g. I can't hear a measurable difference). But I can with 100% accuracy pick 128 kbps mp3 from wav. Lots of people claim they can hear things that they really can't 🙂🤘
Depends on what quality RUclips decides to send to you. But the artifacts captured in the nulls and delta will be audible regardless of the codec used when watching the video. But to answer your question, it was recorded at 48 khz / 24 bit. Rendered down to 16 bit for uploading to RUclips.
The largest problem with audio quality is arguably loudness war, essentially compressing the audio so that it sounds louder on the radio (and attracts more attention, to become more popular and sell more (make more money)). It's a disgusting practice where someone decided to ruin art because of greed. I would take 16-bit 320 kbps Opus with high dynamic range and genuine mastering for audio quality rather than a conformance to the status quo of loudness war, over a 24-bit lossless FLAC with loudness mastering practices. I am unable to listen to a certain Metallica album for this reason (you know, the one infamous for sounding bad). You wouldn't take the real Mona Lisa and scan it into a JPEG file and then print it out in double the resolution on a state of the art printer. You'd never be able to get back the information lost when compressing it, the printer would just print out all the artifacts and each 2x2 pixels would be copies, providing no additional detail or difference. Music having a low dynamic range (and the additional artifacts like distortion and noise) also makes it harder to hear the difference between codecs. This is not a good thing, it just proves how degrading it is. It's such a facepalm for streaming services to even suggest 24-bit sound when they still haven't fixed the mastering practices. The point of this is also for people who have proper audio gear, not for earpod users... Most of these side-by-side blind comparisons tests are commonly based on heavily compressed audio with low dynamic range. :/ Loudness war has been inaffective for a long time because of normalization. Radio also isn't very relevant, and just like commercials being super loud people instead automatically mute the sound when it begins. Loudness war is like a very addictive drug quickly gaining tolerance. While the video is interesting, it's kind of an impossible task because of RUclips's compression. The differences demonstrated in this video are lessened and some aspect possibly removed (as it adds in whatever you're supposed to not hear in the uncompressed version like noise floor and artifacts). It's kind of like showing a uncompressed 4k footage with x265 and AV1 encoding and then convert it all to 1080p VP9. If the bitrate of the compressed examples is low enough you can still see a difference, but you would never see the original uncompressed, even on a high-end 4k monitor (analogous to having a 24-bit quality DAC and good headphones to listen to this video). As for the analogy used, a 1080p video file can look much better than a 2160p video file, depending on bitrate, format etc. There are so many steps that can lower the audio quality, it all starts with production.
I get it still.. The scientific visual difference between the digital representation and the analog representation is great. Drop a pebble in a lake you get a perfect balance circle resounding out to the banks and back. The digital representation is a jagged edged circle. Sliced so thin that it’s not noticeable to the conscious observation actions . How ever the biological response is Baron in comparison. Inflection presence present in the same fullness from progenitor is not there. The biological reaction to stimulus as well as the psychological response differs. It is what it is…. Argue Kinda like vitamins.. the identified active ingredient is there but all the wisdom is left behind…
You have actually demonstrated just how very audible and horrible compression actually is. Unacceptable. And I was one who hopefully wrote of those algorithms in the 90s. How very wrong I was. It's bad, it's audible at worst and "feel able" at best. Lossless only for me.
Isn't that you are showing the DIFFerence between master and encoded material which means everything what was hear-able that was "the noise" which was CUT-OUT from the original material to the encoded file? I don't know this software, but the setup on the screen is actual to listen the DIFFerence. When you add that as you say that "noise" back to encoded file, the result should be master signal (the source). So it is not adding any crap to the output, it just removes the "noise" which human ear can't hear or it is degrading high frequencies (cuts out) in lower bit-rates (so that is why you hear more "noise" on lower bitrates which works as intended). The diff comparison can also be easily achieved using an Audacity or other software by comparing tracks with one having inverted phase. But there was some poor codecs in the past like Blade which added some squealing artifacts to the output but this is a different story.
Yes you are right - this is the DIFFERENCE between the wav and the lossy file formats. You can perform this yourself by encoding different file formats and then inverting the polarity. That is how this software works. Like you mention, it's a little unclear if the difference signal has been ADDED or is MISSING from the encoded lossy file. Regardless, its a great example of what the sonic differences are. Lots of people think lossy formats change the sound MUCH more than this. I'm quite skeptical after doing the blind shootouts :) Thanks for the comment!
@@RaytownProductions thank you for your response, and of course the video itself - because I was a bit rude at the beginning and need to correct myself :-) I'm not a musician but I'm interested in lossy music compression formats just from the practical stand point of view. I have also found another great material about that subject: ruclips.net/video/Nrdth-Bw1B0/видео.html (I think it has English subtitles). So Mr Adam did a lot of comparisons (video 33 and video 34) of different quality setups and I was actually amazed how little difference is between original track and HQ (Extreme and Insane) Lame MP3 presets those days. The material is worth watching also like yours because it is very educational and contain some outstanding quality knowledge. My conclusion after doing some research also is that above 192 kbps MP3 and AAC are non distinguishable between them, and the LAME extreme is actually excellent quality which is at least for me sufficient for HQ compression. For car (if your data space is crucial) of course we can go as low as 128kbps because there is a lot of background noise anyway and for headphones I would just stick with extreme always. And I always stick to MP3 at least for now (and FLAC for lossless) because of the best compatibility and availability - just my few cents addition to that.
very short tldr almost no difference as far as i can tell very long tldr difference mode: mp3 320 adds/removes some sort of noise (like pink/white/other color noise) and more so when the base hits mp3 192 bout the same as mp3 320 but a bit more extreme mp3 128 same as mp3 320/192 but way stronger (IMPORTANT: apparently what most streaming services use (aka youtube, spotify, ect)) aac lc 320 very slightly less dramatic than mp3 320 aac lc 128 a good bit less dramatic than mp3 transcode mode (comparing og audio to compressed audio directly): IMPORTANT note that audio may be affected by youtube compression mp3 320 no audible difference as far as i can tell mp3 192 still none mp3 128 yea either i have bad ears or i cant hear any difference aac lc 320 maybe a little worse quality than no compression aac lc 128 no audible difference that i can hear between this and aac lc 320 OR mp3 128
You sho up d get yourself some old school vinyl set up. With say a class A amplifier some classic cabinets a quality turntable and diamond cartridge. Ha. And you will laugh at those that pontificate about digital greatness of sound reproduction. Laugh Out Loud thank you for reading . Too bad that before we know it.. this quality of sound wave reproductions will be lost and will never touch billions of ears. Even live music is mutated on going as a practice. No? Sensitivity gets sliced out. And we become what we steep in. It is what it is. Ha and Look already I’m the only one saying this. Laugh Out Loud yep. Hey not many people know about A-432 orchestral/ instruments tuning either…
You are correct. They have stripped the ultrasonics from MP3 and flac. It's like stripping the soul out of the music. Vinyl still has ultrasonic frequencies which is why vinyl is more pleasurable to listen to. Sad that most people don't know this and don't know what they're missing.
I don't think that is necessarily true. A very simple experiment you can do to prove this is to export a WAV file at -0.2 dbFS then do the same thing but export an mp3 at 128 kbps. When you open both files up, the mp3 will likely be clipping due to errors in the encoding process (at least with broadband signals that are at competitive levels like commercial releases). Those errors result in an increasing signal that drives your mp3 file into clipping - that is to say, it is ADDING something to the signal. Usually this is noise or distortion artifacts from the compression algorithm. I'm not a DSP expert by any means, so your mileage may vary, but this simple experiment doesn't jive with your theory. Digital signal processing can be really confusing 🙂
Personally I always stick with wav files, rather than any other codec. Wav can be copied over and over without any noticeable loss. As in create a copy of the original and then a copy from the copy and so on. Try that with any lossy codec such as mp3 and you will eventually end up with a heavily degraded audio file. But if it is for streaming, you can get away with it, but I prefer not. One simple question. This is a video on RUclips which is highlighting the differences between these codecs, but we're not actuially listening to any at their ultimate, as was stated at the beginning, any such video is likely to have it's audio converted to mp3! And then you have the fact that most of todays population are so used to listening to the sound of mp3, that it really makes no difference as to the way in which the audio is encoded because their ears have become used to listening to an inferior sound. Just as their ears have become used to the over compression due to the loudness wars. In fact they are more likely to complain if there just so happens to any dynamic range. Sadly remixes of audio that previously had great dynamic range, especially that which was originally recorded in analogue is being ee-released and is now noticeably compressed. I hate it, especially with the latest Elvis Presley re-releases. They've been re-released without the overdubbing imposed on them by Elvis' producer Felton Jarvis who should have left them the way that Elvis had recorded them. Which was the way he wanted them to sound. But what has ruined otherwise fabulous releases, is the over compression. To such an extent, you can actually hear Elvis' voice being pushed back into the mix. Worse still, in unofficial releases, it disappears. I wish that fans wouldn't buy these as they cost an arm and a leg compared to regular releases and just because it sounds crisper and louder doesn't mean better.
I appreciate your effort but i found your clip not correct because the moment you upload this video to RUclips all we can hear is just an AAC 320 kbps Format in best case scenario!! SURPRISE you can't upload mp3 or WAV on RUclips so how can we tell the difference that you talked about? this is not your fault... this belongs to super poor quality audio which RUclips deliver to any video on their Platform... so please don't mislead the people and keep the good and the correct infos up ...Cheers
I show the difference between the mp3 and the other formats explicitly by subtracting the uncompressed from the compressed formats. So regardless of the compression being applied by RUclips, you can still get a sense of the differences. I know it's convenient to skip around a video, but if you have time, try to watch the entire video 🙂
I havent watched the video yet, but mp3s sound pretty bad, imo. They reduce the highs, smear the bass, and the audio loses detail - even for single instruments or vocals. AAC also sounds not great. OGG sounds a bit better, because it has a more present top end, but still not as good as lossless, like FLAC, WAVPACK, ALAC and WAV. My 2c. :)
VINYL ANALOG RECORDING IS MILES AHEAD BETTER THAN CD OR ANY DIGITAL LOSSLESS FORMAT not even close. I have the collection of Vinyl like Christmas song by Ray Conniff, Michael Jackson Thriller ang many others. I have both of these in Vinyl Analog Recording & CD. OMG, the treble, bass, and all the mids from vinyl are full blast!!! Just compare them in actual operation head to head.
This is true. It's because flac (CD) and MP3 have most of the ultrasonic frequencies stripped out of them. Vinyl has a lot of ultrasonics still on there, so it sounds more natural and should be more pleasurable to listen to. But there are ways to add the ultrasonics back into flac and MP3. Once you have done that then it should be just as good as vinyl. It's a shame most people don't understand this and are happy to listen to their flawed music in ignorance.
@@RaytownProductions Analog vinyl recording not digital recording. Better back read. VINYL IS MILES AHEAD BETTER THAN CD OR ANY DIGITAL LOSSLESS FORMAT. I have the collection of Christmas song by Ray Conniff, Michael Jackson Thriller ang many others. I have both of these in Vinyl Analog Recording & CD. OMG, the treble, bass, and all the mids from vinyl are full blast!!! Just compare them in actual operation head to head.
@@narvaeztiamson918 I think he meant the source before pressing. Because almost every modern recording & modern vinyl are digital recording, digital mixing. Unless it's an old recording (analog master tape before the late 90s) & an older vinyl (analog mastering).
![96 kHz vs 48 kHz vs 44 kHz - What's (really) the Best Sample Rate for Audio? [2023]](http://i.ytimg.com/vi/g0BpVO16dbI/mqdefault.jpg)
I've had people asking me for FLAC files of a project I recently did, and I didn't really know why. I've had no problems with MP3's for many years. So I exported a full FLAC file, like no compression, the file is 200mb, it's insane, and A/B tested it against the normal 8mb 192kbps MP3 file in Studio One. I could not tell a difference on my monitoring headphones, whether knowing which was which or not. I'll still do the FLAC files for other people if they really want it, but compressed files sound perfectly fine, to an extent. I know how terrible low quality MP3's are, I used to have many of them, and some of them are still kicking around in my library since they're very rare songs. Bumping up to 320kbps doesn't really change the file size for MP3, so I've moved up to that, but I'm still going to listen to MP3's and be content.
Nice. Thanks.
Thank you for the video. Because the encoders are very good these days, it is generally really difficult to hear, especially in higher bitrates. But keep in mind, the audio codec here on RUclips is already lossy, so we can't never hear the real uncompressed WAV audio in this video.
Perhaps, but we are hearing the difference signal!
LAME 3.99 & newer are pretty much reach transparency at V2(192kbps) I can't tell that from Lossless. But I do encode some edge cases at V1 & V120(Helix MP3).
I did these blind tests a few years ago and I honestly couldn't tell the difference between 320 kbps and the wav file. and when I was just choosing based on what I thought sounded better I was predominantly choosing the MP3. That was very humbling and nudged me down a rabbit hole of digital audio and audio science in general. and its been a great ride! if only more engineers and audiophiles would try these test themselves. its only made me smarter, not to mention has save me some money! great stuff!
The problem is when you export an mp3 file through your DAW it sounds compressed and distorted in a weird way. It makes the song sound cheap
@@DicksonDaggerCould be a problem with your DAW's export feature. If it doesn't use LAME, uses an older, outdated version, and/or is double encoding (re-encoding multiple times) it will degrade sound quality pretty quickly.
I always export from my DAW at lossless and/or with lossless stems, just to be safe. And then I encode later with a tool like Foobar or fre:ac. This maximizes quality, keeps a lossless version of your music for archival purposes, and makes it easier to share the mp3 with others!
@@TheDylandProductions My point is your track is going to sound better because you started with lossless files to begin with. But when you start processing lossy files like mp3, that file gets recompressed again if you export as mp3 and you risk losing quality. I’ve had the my mixes sound great in my DAW using mp3 beats but once I export it as mp3 the signal distorts in a weird but subtle way.
Apple did not invent AAC, they simply used it along with other manufacturer.
RUclips down samples audio to at best.... 160kbs (Opus 251). That is what this video's highest audio quality is at its maximum. YT is veryy sad when it comes to throwing away the large majority of all the original data the video maker wanted.
I asked my producer friend (he produces electronic music, and I mostly listen to this kind of music so i know what it's supposed to sound like) to export the same track directly from the DAW in a couple different formats (@ 44100 Hz, since higher sample rate =/= higher quality), and we listened to them in parallel firstly blind, and afterwards with knowing which are which, and here are our impressions:
Lossless formats (WAV, FLAC & AIFF, 24-bit and 16-bit each) - We could pick out the lossless formats in the blind test, and we agreed on linking the sound quality of these files the most, but we couldn't tell them apart from one another, even after multiple listens... 16-bit FLAC-s have the smallest file size, and plenty of tagging capabilities, so this would be my pick for Hi-Fi listening, while sound engineers, producers and DJ-s should go for WAV for compatibility and editing reasons.
Lossy formats (MP3 & AAC) - We could tell the lower quality of these files, but here I could also place the lossy files into 3 tiers of quality between themselves, while my friend just said "these all sound terrible", and stopped listening 😅.
-1st tier is 320kbps MP3-s with CBR and "Highest quality" MP3-s with VBR: I couldn't tell the difference with these 2, but the latter has slightly smaller file sizes. These 2 are perfectly fine for entry level Hi-Fi (eg. $20-$50 IEM-s or speakers), and is what I use for my library since the file size diffrences between this and lossless are about 60%, with the quality staying really close.
-2nd tier are 320kbps AAC & 192kbps CBR: Louder parts sound identical to the first tier, but quieter parts of songs is where you can tell there is information loss. These are a hefty 80% smaller than lossless and are also fine for average listeners, but I prefer HQ MP3-s.
-3rd tier are everything else (128 CBR, 192 & 128 AAC and MQ & LQ VBR): All of these don't sound good to my ears, although the difference is harder to notice once I tried some lower quality earbuds. I'd stay away from these, unless you're desperate for storage space.
While this tier list might suggest otherwise, the diffrences between all of these weren't THAT great, and to someone who didn't take this kind of critical approach to listening, they might have been indistinguishable (except the 3rd tier, where I think everybody would be able to tell). Also keep in mind that this test is done on electronic music where audio is pushed to it's limits compared to other genres, and I imagine the diffrences would be even harder to tell for pop or classical music.
PS English isn't my first language so forgive if I have some spelling or grammar errors.
Great analysis and suggestions! Thanks so much for the comment 🙂
Even on a 500€ HiFi, the LAME V0 sound fantastic. Saying a transparent (MP3 > 200 kBit/s) sound "Terrible" just shows he is talking out of his butt.
The digital equivalent of analogue sources 30+ years ago : good quality = cassette / excellent quality = Vinyl LP / Studio quality = Reel tape (at least 15ips), but most consumers then listened to cassette through headphones so enjoyed the worst version of the master, so the majority of music consumers taking streaming now via spotify etc is getting a pretty decent sound if you compare it to the mass consumption of the cassette back then.
Initially, in the early cassette releases, but as they became more main stream, the quality massively improved. I'd prefer a high quality cassette release to mp3 any day. The only downside to a high quality cassette was the tape hiss, but then mp3 carries with it, it's own inherent noise. As for vinyl, it used to be the best, but as with cassette, its sound was degraded overtime. It increasingly crackled. The best sound is still CD when the audio is AAD, rather than DDD as you then get the best of both worlds. Providing that is the record lable isn't lazy and simply re-releases the original vinyl master.
Cassette is better than flac and MP3 because it has ultrasonic frequencies. MP3 and flac have nearly all the ultrasonic frequencies stripped out of them. It's why vinyl is still popular.
@@dtz1000It's not the only reason why vinyl is popular. Lol
I only notice a difference when the mp3 is a low quality conversion, and/Or has artifacts.
Well done. I've always been able to hear the artifacts, and this demonstration really brings it home. I agree that most people listening to streaming in their car, or on cheap audio gear will likely never really notice, especially without the A/B comparison.
When working on a mix, or listening on my high end stereo equipment, the MPS really bugs me. I have a client who sent me his tracks recorded with MP3, and I just could not make them sound right. I made him render to WAV. After he did, I had him A/B and he could definitely tell.
Lastly, i heard some pumping in the left channel cymbals. It's there is the WAV, but it got exaggerated when compressed.
Anyway, great video. Thanks!
Yeah, lossless like mp3 really bug me too. :D
Great video. Very interesting. I have done some listening tests between AAC and Apples's new Lossless formats. Only the Max 24 bit 48 KHz lossless stream through the Apple donegal D/A to my Sennheiser and RODE headphones. I can't tell the difference even though I can still hear up to 12KHz (I could hear up to 17 KHz in my 20's).The AAC compression noise (on your video demonstration) sounds higher in frequency than the MP3 noise. Defiantly better I think especially for my 70 year old ears.
Watched most of tutorials and few full mixing mastering videos in 4 days :D Thank you Sir!
There is an ABX plugin for Foobar2000. You can encode files at different rates, then test yourself. In Archimagio's 320K MP3 test people could not tell the difference.
The AAC codec should work better than MP3. Both codecs were invented by the Fraunhofer Institute. MP3 came first and AAC was created to overcome deficiencies in the MP3 codec. Given your examples, I believe they succeeded.
I dont think it is adding noise, the noise is the sound of the difference. you can still kinda hear the vocal in there even on the high quality. I think it is total cancelation, like a null test, except where lossy format has removed some information.
Yep! That's exactly right. It's the DIFFERENCE between the original and encoded file.
If it’s the delta, then what you are hearing is the audio information that is being discarded by the compression.
@@RaytownProductions so the seamonkeys are really the result of the abrubt subtraction, especially noticeable in transients
Someone actually showed what the real difference is. At least now I know what it is.
I like listening to early 2000s music like Staind in mp3 because that’s how I used to listen to it when I was growing up. Either that or Flac (CD quality.) but for 2010s music I use RUclips and use the original videos that I listened to lol. Because that’s why I did in highschool. The artifacts add to the nostalgia for sure. Its like tape distortion on an old record or crackle. When I listen to 30s music I use vinyl rips. Always use wired headphones or earbuds. Cuz that’s what I used to do. Lol maybe I’m crazy but it helps with my anxiety and producers even use this technique in their samples to create certain a certain timbre. I think people need to realize Bluetooth is ruining their music way more than these lossy compressions because it is also lossy compression but in real time. Which is much harder and causes audible artifacts that don’t sound good really. Unless Ofcourse you grew up listening to that 😂
I used to collect or create my cd albums as FLAC files, but now I have a free disk space thanks to you 😉. Since listening to all of those different LOSSY formats didn't make so much difference why keep FLAC files, right? Cheers
just remember to convert it at highest sample rates as possible (320Kbps for MP3) and you're pretty much good to go.
The difference is VERY subtle, the aac sounded a tiny bit clearer vs mp3 which sounded a bit more noise-y to me.
IDK if more noise makes for better quality or vice versa. Thanks for the info!
Are you sure that's the noise that's being added? I would assume what you hear is the delta and not what is added. Meaning what you hear is what is missing, not what is added. But I don't have this plugin so I don't know if it's actually what is added but seems a bit strange.
You are correct. It's what's being removed, not added. He is saying it wrong. Really should clarify that in the video
I could hear a slight difference. The punch and/or depth seem to be noticeable when using the mp3 or aac. But, I'd love to get the Fraunhofer and do maybe more comparison with different types of music as well. I think some codecs maybe better for certain styles or genres of music. With me getting into doing more mixes/mastering, I'm wondering if I can find a way to mix/master for the betterment of these codecs and etc. All in all, it may not even be worth the time to invest, as different sites offer different lossy audio anyways, it is what it is, lol. I think I was just disappointed noticing a Video mix I did to an AAC format, I could hear my own difference, and was concerned. So, it brought me around to this video, lolol! Thanks for the video!
So what you're saying is, Opus performs better than both MP3 and AAC.
Ok, my previous question really doesn't is a problem for the video itself afterall, like I thought anyway. Great video, very eye-opening. Thanks.
What I noticed is that AAC LC stayed closer to WAV. it slightly brought the vocals and kicks a bit more forward but not by much. the MP3 files heavily brought forward the Vocals and Kick drums and made them sound more up front and centre, the track became slightly unbalanced and lost it's space and depth, some frequencies clouded over others.
For me, the best audio for both experience and size is 256kbps aac or 320kbps aac 48000hz
Thank you for the video in was very insightful!
I know there is a difference in sound quality, but I don't really hear it unless I focus very hard on it. I keep using mp3 320kbps most of the time which is a good enough quality for me, I feel like mp3's have a slightly punchier bass which I love, especially over aac which sounds to have a bit less bass (could also be just in my head lol). Personally I would only go with FLAC if I had a crazy good expensive audio setup which I and most people probably don't have. Good video, thanks.
MP3 on 320kbps is enough for most sound systems and ears
320 is not bad but as an engineer I can hear the artifacting in 128 and its impossible to ignore. For just listening who cares? most people cant really hear it because the dont know what they are listening for. To them it's part of the song.
You have your information backwards. The program you're using is doing a null between the two audio codecs, so the noise you are hearing during the null is the audio being removed during the compression NOT added. The difference between the two tracks.
I'm using mp3 for format which is the 320kbps for a long time, however the website that I'm using was down.
Note: ALL digital audio formats are compressed, WAV FLAC included, unless you have the band playing live in your livingroom.
Modern music sounds so bad compression is the least of your worries.
mp3 can go much higher than 320k, ive heard 680k mp3 and it sounds superb
The differences between WAV and mp3 below 320k are that the stereo separation gets bitcrushed.
I don't think it's noise being "added" through the conversion but rather a lot of that noise is what is taken away, hence hearing the difference
p.s.: the "seamonkeys sounds" being present (once listening to the mp3 itself and not in difference mode) could be the same modulating pump of that noise taken out leaving you hear more high mid-range pumping weirdly
This is a good point. We don't know if it's added or taken away based on what I showed. I misspoke. Thanks for clarifying!
@@RaytownProductions For what it's worth, I often had that lingering question every time I do a null test: is it something added, subtracted, or both. I wish there were a clever way to know (or maybe I just didn't think of it yet)
Very interesting. thanks Bobby.
A lot of what you’re hearing isn’t necessarily a reduction in transients, but a changing of them. Lossy audio codecs have to quantize time information to an extent. Sometimes there are very slight “pre echo” artifacts that result from this, and will show up as the slight “transient smearing” effects that you’re describing when you’re trying to do a difference between files.
The reason that this happens is technical. It has to do with MDCT windows, which are used to encode frequencies present in given time frames/“windows” that are divided out from the original file. These aren’t perfectly precise with time accuracy, the smallest windows available are 128 samples on AAC or 192 samples on MP3, which means that it will never be able to encode every transient with sample-for-sample time accuracy (even if they’re not really “losing” much information in terms of frequency content).
It’s just a trade off as a result of how these lossy codecs work. There isn’t really a whole lot that can be done about it aside from using different methods to compress, but MDCT transform algorithms (MP3, AAC, Ogg Vorbis, Opus, etc) are pretty effective compared to other approaches for compression.
I feel like this track (although it's killer btw) is a poor example to use for something like this. It already has some distortion and lo-fi sounding elements.
Great video, I been uploading music mixes on RUclips and always wondered which would be the best sound recording aac wav MP3 etc etc, I'm still kinda undecided which is best wav and aac, but your explanation was great.
Yesterday, as it happens, I was listening to CSN&Y's Deja Vu album, which I had ripped to MP3 a few years ago. I, usually, listen to the CD and had not heard the rip for a long time. I heard some nasty crap, coming out of the speakers, and was pretty shocked. I re-ripped it to FLAC and it was completely cleaned up. So, was it the change in CODEC or was it just a crappy rip to MP3.
Ironically, I have been in the process of re-ripping a lot of my music to FLAC, as I feel that it is, significantly, better. I, also, tried WAV but, in truth, could hear no difference and WAV takes about 5 times longer to rip, so FLACit is.
It might be my shitty earbuds but I actually kinda liked the 192kbps conversion better than the uncompressed. It kinda gives it an analog sort of warmness.
There is a fine line between objective and subjective sound quality. There are many variables, such as headphone quality, a person's age, environment, etc. Remember the Yanny vs Laurel argument from a few years ago? About half of the people heard one sound and half the other, and it suggests that everyone hears slightly differently. Back in the early 2000s when disc space was a premium, I encoded all of my music in wma format @44100hz 192kbits and now I'm stuck with it, unless I reencode everything. The last few years I have used FLAC, and that seems to be good enough for me. Interesting and fascinating video.
Not only do we hear differently but we are used to different sound signatures. We have a different idea of how an instrument is supposed to sound and how songs are supposed to sound. It even takes a little while to get used to new audio gear, which makes it hard to discern if it's different in a good way or a bad way.
If the files we use have issues, like the ones cause by loudness war (low dynamic range, distortion, artifacts etc.), then that creates a bottleneck and a possible distraction. Perhaps better audio gear makes it easier to hear certain artifacts? That way the blame could be put in the wrong place.
We are subjective by nature and since we haven't heard our playlists in the studio and on the gear used to produce it, our imagination can only go so far and as a result we are susceptible to biases and subjective experiences. Especially since different ears don't hear the same thing from the same audio setup.
Blind audio comparison tests also suffer from these issues. If you pay attention to the volume, it can make you perceive better quality. If you listen to high hats, when the difference is in other places, then you'll be similarly distracted. So knowing what to listen for is not a bad thing, especially not when you consider how easily it is to make tests where you won't hear a difference between 128 kbps and lossless FLAC. For instance using audio files heavily compressed from the studio that sound bad to begin with and of course using low-quality audio gear. But make sure there aren't any bottlenecks in this long and complex chain, you'll definitely hear a difference even if you don't know what to listen to.
Compression encoding also alters the frequency range and its algorithm can prioritize certain aspects. I find it interesting to consider the implications of the algorithm's removal of noise, grit or other sounds that are located outside of the included spectrum. Some codecs even stops at 12k. I think it's plausible that factors that are unrelated to the lossless format itself can cause false results in codec comparisons. Like I said, if you listen to the wrong details or have the wrong assumptions - you risk being misdirected into incorrect interpretations.
Интересно, кто-нибудь пытался сравнивать сжатый аудиофайл с тем же файлом сжатым повторно ютубом?
In all honesty listening to the track in the different codecs (on a tablet I couldn't hear a difference) I will check it out on my various monitors to see if I can hear any differences. I would like to make one point though, how many music consumers listen on ear buds or less than optimum speakers and in less than ideal environments? My guess is most. It is a sad shame that audio mixers/producers put so much effort and time into trying to create good or even great sounding music, only to have the online music streaming services slap their work through these codecs.
While AAC was made to replace MP3 in 2003 the open-source community worked on the LAME MP3 encoder all the way up until 2017. Long after the Fraunhofer Society abandoned the format for AAC. It absolutely guts the original 1998 Fraunhofer codec and AAC in all but the lowest bitrates. It's pretty much been universally adopted in every program that uses MP3 these days. Even Fraunhofer uses it over their own codec. In many ways, it's an obsolete format but it was polished to absolute perfection by dedicated fans to remain competitive.
I'm not sure that is the best type of music to judge this on. To me it just sounds like a distorted noise with no dynamic range. Try a REAL bass and drums and a piano maybe. Don't brick it, chose something with a dynamic range. The differences will show up much more.
How much audio bit rate did you use when you uploaded your video? It sounds terrible even for the lossless!
What the codecs basically do is slice up the sound into segments and reduce the signal to noise ratio in each such that it follows the overall level per frequency band and you always get roughly X dB of signal. Whether something gets added or taken away is a matter of semantics. Add something with a negative sign. Negative birth as a euphemism for death. The codec may zero out a band completely, taking it away, but over a human perceivable time scale it's still a burst of noise.
Most people listen to compressed audio without having a choice, on the radio, in video games, over wireless headphones. 128 kbit/s is really bad and the sound gets worse every time it gets reprocessed, such as when the radio station loads it, transmits it over their lines, then it gets into headphones or when people encode it for their collection. Even if it sounds acceptable after one generation, it's better to start with a better copy.
Thanks man.That is a good info.
This is really really useful info...but RUclips won this battle--the data is there, we just cannot hear it. I think either high-quality headphones or being in-studio with you would be the best way to hear it. I'm an audiophile and I could barely make out a difference, however this is great to know!
Who is Best for music MP3 320 kbps vs m4a (AAC) 256 kbps?
good song! yours?
The point is you don't just listen with your hears, your whole body listen and your cells are catching the frequencies you can't hear. Check bruce lipton biology of beliefs.
What you meant right after the 3 minute mark that streaming audio services use 128kbps MP3, wouldn't it be AAC? Was it a mistake? I mean no way stuff would sound so good with 128kbps MP3. Only if back in the day when I listened to 192kbps MP3 they were very bad rips, because until something like 256kbps I can definitely notice problems with MP3. Anyways, going to watch the rest of the video, seems pretty cool regardless of my question.
Thanks alot for the test! Could I ask on which encoder you used to make the aac?
He's using the Fraunhofer encoder as a part of the plugin called Sonnox Fraunhofer Pro-Codec Plug-In. Don't ask me what Sonnox is, but the encoder is Fraunhofer. ;)
The codec used is Fraunhofer AAC-LC with CBR (Constant Bit-Rate). Not all encoding parameters are visible.
Different companies have their own AAC encoders and you don't need this program for encoding, if that's what you wonder. This plugin is just an interface for the Fraunhofer encoder.
@@gung-hochang9573 thank you, i wish you the best
MP3 is removing things we can't hear - by law of physics or by our ears or by your equipment. The people at Fraunhofer were scientists, they knew what they were doing. People crying about MP3 "removing things" are surely also crying about their BluRay not containing an infrared video track. You could not see it, but hey, "at least it's there, for the peace of mind".
I have read extensively into MP3 and other lossy codecs and they should be complemented on how great and clever they work instead of just "zipping" the raw data. It's like removing empty pages in a notebook and only carry the written pages - I prefer that to having GBytes of data information that we BARELY can hear.
You are correct that the ultrasonics are stripped out of MP3 and even flac, but you are wrong when you say it doesn't matter. Those engineers were stupid, not clever to do that. It's why vinyl is making a comeback as it contains a lot of ultrasonics.
Really the wrong choice of music for trying to hear differences. Something with less guitar distortion and unplugged would have been best.
Whats the title of the backgroundmusic?
Thank you, but every time I listen to this song, it gets me so excited that I forget about the content of the video🤣🤣.
is the difference heard in this software taking in consideration psychoacoustic auditory masking or is it just a substraction of the 2 signals ? if it is just a substraction this is not very usefull as it will mostly expose what a human ear cannot hear :)
You need to know what to hear to hear it
couldnt tell the difference between any of them
when i take a free online blind test of uncompressed vs compressed audio i cannot hear any difference is it my ear limitation ? my ear training ? my gear that is not resolving enough ?
Depends! I have a very hard time picking out 320 kbps from wav. In fact for me it's statistically random (e.g. I can't hear a measurable difference). But I can with 100% accuracy pick 128 kbps mp3 from wav. Lots of people claim they can hear things that they really can't 🙂🤘
Can you say how many times people have said it?
I always can tell 128 if my brain thinks the track was downloaded on limeWire lol
hahahaha
The real question is what is the format for this video's audio?
Depends on what quality RUclips decides to send to you. But the artifacts captured in the nulls and delta will be audible regardless of the codec used when watching the video.
But to answer your question, it was recorded at 48 khz / 24 bit. Rendered down to 16 bit for uploading to RUclips.
Only one sample track does not give full 360 degree view about the differences.
The largest problem with audio quality is arguably loudness war, essentially compressing the audio so that it sounds louder on the radio (and attracts more attention, to become more popular and sell more (make more money)). It's a disgusting practice where someone decided to ruin art because of greed.
I would take 16-bit 320 kbps Opus with high dynamic range and genuine mastering for audio quality rather than a conformance to the status quo of loudness war, over a 24-bit lossless FLAC with loudness mastering practices. I am unable to listen to a certain Metallica album for this reason (you know, the one infamous for sounding bad).
You wouldn't take the real Mona Lisa and scan it into a JPEG file and then print it out in double the resolution on a state of the art printer. You'd never be able to get back the information lost when compressing it, the printer would just print out all the artifacts and each 2x2 pixels would be copies, providing no additional detail or difference.
Music having a low dynamic range (and the additional artifacts like distortion and noise) also makes it harder to hear the difference between codecs. This is not a good thing, it just proves how degrading it is. It's such a facepalm for streaming services to even suggest 24-bit sound when they still haven't fixed the mastering practices. The point of this is also for people who have proper audio gear, not for earpod users...
Most of these side-by-side blind comparisons tests are commonly based on heavily compressed audio with low dynamic range. :/
Loudness war has been inaffective for a long time because of normalization. Radio also isn't very relevant, and just like commercials being super loud people instead automatically mute the sound when it begins. Loudness war is like a very addictive drug quickly gaining tolerance.
While the video is interesting, it's kind of an impossible task because of RUclips's compression. The differences demonstrated in this video are lessened and some aspect possibly removed (as it adds in whatever you're supposed to not hear in the uncompressed version like noise floor and artifacts). It's kind of like showing a uncompressed 4k footage with x265 and AV1 encoding and then convert it all to 1080p VP9. If the bitrate of the compressed examples is low enough you can still see a difference, but you would never see the original uncompressed, even on a high-end 4k monitor (analogous to having a 24-bit quality DAC and good headphones to listen to this video).
As for the analogy used, a 1080p video file can look much better than a 2160p video file, depending on bitrate, format etc. There are so many steps that can lower the audio quality, it all starts with production.
I get it still.. The scientific visual difference between the digital representation and the analog representation is great. Drop a pebble in a lake you get a perfect balance circle resounding out to the banks and back. The digital representation is a jagged edged circle. Sliced so thin that it’s not noticeable to the conscious observation actions . How ever the biological response is Baron in comparison. Inflection presence present in the same fullness from progenitor is not there. The biological reaction to stimulus as well as the psychological response differs. It is what it is…. Argue
Kinda like vitamins.. the identified active ingredient is there but all the wisdom is left behind…
You have actually demonstrated just how very audible and horrible compression actually is. Unacceptable. And I was one who hopefully wrote of those algorithms in the 90s. How very wrong I was. It's bad, it's audible at worst and "feel able" at best. Lossless only for me.
Isn't that you are showing the DIFFerence between master and encoded material which means everything what was hear-able that was "the noise" which was CUT-OUT from the original material to the encoded file? I don't know this software, but the setup on the screen is actual to listen the DIFFerence. When you add that as you say that "noise" back to encoded file, the result should be master signal (the source). So it is not adding any crap to the output, it just removes the "noise" which human ear can't hear or it is degrading high frequencies (cuts out) in lower bit-rates (so that is why you hear more "noise" on lower bitrates which works as intended). The diff comparison can also be easily achieved using an Audacity or other software by comparing tracks with one having inverted phase. But there was some poor codecs in the past like Blade which added some squealing artifacts to the output but this is a different story.
Yes you are right - this is the DIFFERENCE between the wav and the lossy file formats. You can perform this yourself by encoding different file formats and then inverting the polarity. That is how this software works. Like you mention, it's a little unclear if the difference signal has been ADDED or is MISSING from the encoded lossy file. Regardless, its a great example of what the sonic differences are. Lots of people think lossy formats change the sound MUCH more than this. I'm quite skeptical after doing the blind shootouts :)
Thanks for the comment!
@@RaytownProductions thank you for your response, and of course the video itself - because I was a bit rude at the beginning and need to correct myself :-) I'm not a musician but I'm interested in lossy music compression formats just from the practical stand point of view. I have also found another great material about that subject: ruclips.net/video/Nrdth-Bw1B0/видео.html (I think it has English subtitles). So Mr Adam did a lot of comparisons (video 33 and video 34) of different quality setups and I was actually amazed how little difference is between original track and HQ (Extreme and Insane) Lame MP3 presets those days. The material is worth watching also like yours because it is very educational and contain some outstanding quality knowledge. My conclusion after doing some research also is that above 192 kbps MP3 and AAC are non distinguishable between them, and the LAME extreme is actually excellent quality which is at least for me sufficient for HQ compression. For car (if your data space is crucial) of course we can go as low as 128kbps because there is a lot of background noise anyway and for headphones I would just stick with extreme always. And I always stick to MP3 at least for now (and FLAC for lossless) because of the best compatibility and availability - just my few cents addition to that.
My phone is dts and I'm downloading crappy mp4
very short tldr
almost no difference as far as i can tell
very long tldr
difference mode:
mp3 320 adds/removes some sort of noise (like pink/white/other color noise) and more so when the base hits
mp3 192 bout the same as mp3 320 but a bit more extreme
mp3 128 same as mp3 320/192 but way stronger (IMPORTANT: apparently what most streaming services use (aka youtube, spotify, ect))
aac lc 320 very slightly less dramatic than mp3 320
aac lc 128 a good bit less dramatic than mp3
transcode mode (comparing og audio to compressed audio directly):
IMPORTANT note that audio may be affected by youtube compression
mp3 320 no audible difference as far as i can tell
mp3 192 still none
mp3 128 yea either i have bad ears or i cant hear any difference
aac lc 320 maybe a little worse quality than no compression
aac lc 128 no audible difference that i can hear between this and aac lc 320 OR mp3 128
your welcome
You sho up d get yourself some old school vinyl set up. With say a class A amplifier some classic cabinets a quality turntable and diamond cartridge. Ha. And you will laugh at those that pontificate about digital greatness of sound reproduction. Laugh Out Loud thank you for reading . Too bad that before we know it.. this quality of sound wave reproductions will be lost and will never touch billions of ears. Even live music is mutated on going as a practice. No? Sensitivity gets sliced out. And we become what we steep in. It is what it is. Ha and Look already I’m the only one saying this. Laugh Out Loud yep. Hey not many people know about A-432 orchestral/ instruments tuning either…
You are correct. They have stripped the ultrasonics from MP3 and flac. It's like stripping the soul out of the music. Vinyl still has ultrasonic frequencies which is why vinyl is more pleasurable to listen to.
Sad that most people don't know this and don't know what they're missing.
Compression doesn't add anything. It takes it away.
I don't think that is necessarily true. A very simple experiment you can do to prove this is to export a WAV file at -0.2 dbFS then do the same thing but export an mp3 at 128 kbps. When you open both files up, the mp3 will likely be clipping due to errors in the encoding process (at least with broadband signals that are at competitive levels like commercial releases). Those errors result in an increasing signal that drives your mp3 file into clipping - that is to say, it is ADDING something to the signal. Usually this is noise or distortion artifacts from the compression algorithm.
I'm not a DSP expert by any means, so your mileage may vary, but this simple experiment doesn't jive with your theory. Digital signal processing can be really confusing 🙂
Wtf i hear it 👀👀
Personally I always stick with wav files, rather than any other codec. Wav can be copied over and over without any noticeable loss. As in create a copy of the original and then a copy from the copy and so on. Try that with any lossy codec such as mp3 and you will eventually end up with a heavily degraded audio file. But if it is for streaming, you can get away with it, but I prefer not.
One simple question. This is a video on RUclips which is highlighting the differences between these codecs, but we're not actuially listening to any at their ultimate, as was stated at the beginning, any such video is likely to have it's audio converted to mp3!
And then you have the fact that most of todays population are so used to listening to the sound of mp3, that it really makes no difference as to the way in which the audio is encoded because their ears have become used to listening to an inferior sound. Just as their ears have become used to the over compression due to the loudness wars. In fact they are more likely to complain if there just so happens to any dynamic range. Sadly remixes of audio that previously had great dynamic range, especially that which was originally recorded in analogue is being ee-released and is now noticeably compressed.
I hate it, especially with the latest Elvis Presley re-releases. They've been re-released without the overdubbing imposed on them by Elvis' producer Felton Jarvis who should have left them the way that Elvis had recorded them. Which was the way he wanted them to sound. But what has ruined otherwise fabulous releases, is the over compression. To such an extent, you can actually hear Elvis' voice being pushed back into the mix. Worse still, in unofficial releases, it disappears. I wish that fans wouldn't buy these as they cost an arm and a leg compared to regular releases and just because it sounds crisper and louder doesn't mean better.
I appreciate your effort but i found your clip not correct because the moment you upload this video to RUclips all we can hear is just an AAC 320 kbps Format in best case scenario!! SURPRISE you can't upload mp3 or WAV on RUclips so how can we tell the difference that you talked about? this is not your fault... this belongs to super poor quality audio which RUclips deliver to any video on their Platform... so please don't mislead the people and keep the good and the correct infos up ...Cheers
I show the difference between the mp3 and the other formats explicitly by subtracting the uncompressed from the compressed formats. So regardless of the compression being applied by RUclips, you can still get a sense of the differences. I know it's convenient to skip around a video, but if you have time, try to watch the entire video 🙂
lol
I havent watched the video yet, but mp3s sound pretty bad, imo. They reduce the highs, smear the bass, and the audio loses detail - even for single instruments or vocals. AAC also sounds not great. OGG sounds a bit better, because it has a more present top end, but still not as good as lossless, like FLAC, WAVPACK, ALAC and WAV. My 2c. :)
Perhaps you should watch the video before commenting?
Well I was going off the title for the video "mp3 sounds great!" (or something to that effect).
VINYL ANALOG RECORDING IS MILES AHEAD BETTER THAN CD OR ANY DIGITAL LOSSLESS FORMAT not even close. I have the collection of Vinyl like Christmas song by Ray Conniff, Michael Jackson Thriller ang many others. I have both of these in Vinyl Analog Recording & CD. OMG, the treble, bass, and all the mids from vinyl are full blast!!! Just compare them in actual operation head to head.
This is true. It's because flac (CD) and MP3 have most of the ultrasonic frequencies stripped out of them. Vinyl has a lot of ultrasonics still on there, so it sounds more natural and should be more pleasurable to listen to. But there are ways to add the ultrasonics back into flac and MP3. Once you have done that then it should be just as good as vinyl. It's a shame most people don't understand this and are happy to listen to their flawed music in ignorance.
@@dtz1000 real facts!
@@dtz1000how do you add ultrasonics to a digital mix/master ?
Analog Vinyl recordings will only turn all the digital recording into the dust including the lossless digital audio, not even close
Do you mean recording directly to vinyl medium? That is very rare. Most vinyl are made from a digital recording 😁
@@RaytownProductions Analog vinyl recording not digital recording. Better back read. VINYL IS MILES AHEAD BETTER THAN CD OR ANY DIGITAL LOSSLESS FORMAT. I have the collection of Christmas song by Ray Conniff, Michael Jackson Thriller ang many others. I have both of these in Vinyl Analog Recording & CD. OMG, the treble, bass, and all the mids from vinyl are full blast!!! Just compare them in actual operation head to head.
@@narvaeztiamson918 I think he meant the source before pressing. Because almost every modern recording & modern vinyl are digital recording, digital mixing. Unless it's an old recording (analog master tape before the late 90s) & an older vinyl (analog mastering).