AAC vs. MP3: What’s the Best Lossy File Type?
Vložit
- čas přidán 28. 10. 2020
- Learn more about Sage Audio here: www.sageaudio.com
If you’re new to Sage Audio, we’ve been providing industry-leading audio engineering services and education for over two decades and created this channel to help you make professional songs.
![Simple Guide To Music File Formats [For DJs] 🔊](http://i.ytimg.com/vi/4dFgg69RKtc/mqdefault.jpg)
Obviously we cant tell by listening to a youtube video that is compressed. I agree with your conclusion. I always thought 320 MP3 was "cleaner" than AAC, but I attributed the difference to DACs and filters of devices and settings. In fact MP3 is sometimes preferable to the raw audio, because recordings and masters are often dirty or are showing some problems and harshness in some frequencies. The quality MP3 can clean some of the problems on some tracks. My music library is mostly lossless, but for some tracks I keep only a good MP3 that I have chosen over the lossless one. About AAC I think I was always sensitive to the problems of AAC and I was never able to like one. For me mp3 sounds cleaner (it's not always a good thing) but it succeed in removing some unwanted noises and harshness hidden in voices and other air sounds. Also I prefer true stereo than join stereo for imaging reasons, as I can hear the blurred instrument placement when listening to joint stereo. I don't care about loosing the extreme treble because with ageing, I will soon loose the hearing of those.
Thanks for sharing your thoughts! I never thought about it that way - you're right, some people may enjoy the sound of those frequencies being attenuated over having them present. Thanks!
SageAudio.com
@@sageaudio I see similar results as good mp3 with OPUS format. The defects that are attenuated or removed are not frequencies, those are harshness and grainy textures in recordings, for example in voice, or in high hat or snares, or higher pitched noise or saturated signals that have been reduced in mixing but not cut. those are in a broad band from 2K to 8K in general. No frequency is missing when compressing audio with enough bitrate. The filtering and missing frequencies happens when bitrate is reduced too much. There is a compression level witch is qualified as transparent by a jury of careful listeners. Transparency threshold is achieved when in double blind tests the listeners can not tell difference between lossless and the lossy compressed file. The reason I can hear differences where the jury could not is because the issue occurs on SOME TRACKS and not in other tracks. This is specific, not in general.
Qual codificador de mp3 você usa ?
I can’t believe I actually heard the difference! I think it’s because you showed us what the sound being removed sounded like first. I picked up on that quickly during the test. On other tests like this I never know what I should be listening for. I also have noticed that on tests like this, that it’s the examples with crispy percussion like high hats or snappy snares that I can usually guess correctly. That kind of makes sense now. It seemed like the lossy formats were removing transients. Nothing tonal really came through on the null test. Awesome video. I’m so happy I found this channel. Love the topics and the format!
Hey David! Great to hear that you could hear a difference! I think lossy files become the most easy to notice when there are a lot of higher frequencies like you said. Thanks for watching - glad you're enjoying the channel!
SageAudio.com
I nailed it. But only because you explain that the AAC was the worst quality. My ear heard the difference. Crazy
To everyone who's commenting on the "blind test" and how they were able to hear the difference: you realize that you're listening to a 160kbps OPUS audio, right? Surely, you must know that every CZcams video has 130-160 kbps AAC/OPUS audio track.
Thanks for watching!
SageAudio.com
thankyou. This was bothering me throughout the results section
Thank god! I was feeling bad because I thought the AAC was better...
I guessed correctly the WAV file only and got the other two switched. I think that if you know what is cut and have a trained ear, guessing the WAV file should be simple.
On a final note, I really appreciate how you approach technical questions. Your videos are outstanding.
Hey Synaptic Schism! Thanks for watching this one! Cool to hear that you and other viewers could tell which was the WAV, but the lossy format was harder to tell which was which.
Thanks for your kind words about our videos! We appreciate you taking the time to watch them
SageAudio.com
it's simple anyway the quality wasn't as high in play back ;)
I immediately thought number two was wav, couldn't tell which was which with the others. I really think it was a volume issue though and that you may have cut in on a peak on the first AB. Since my hearing is down 10db from 20Hz to 1k and plummets to 70db down by 3k that's about all I can imagine I was hearing. Nice test though. Good info and yes, very surprising.
And for those that listen at high volumes for ANY length of time - WISE UP! You do not want this. You can still enjoy music in my state, a lot really, but you know there is so much that is just no longer there. Cymbals, harmonics, whispers in vocals, etc. Your ears are not like your eyes. You protect your eyes at all times because you know that blindness can happen in a split second. Ears go slowly and you don't notice. I'm 65 and own lots of expensive great gear but it's really lost on me. Too bad. Loud music, power tools, loud environments, etc. Avoid them all. Three bucks for cheap but perfectly good ear protectors from Harbor Freight would have saved me. Picking up a hammer? Put the ear muffs on first. Always. Again, don't be the young invincible 'it won't happen to me' kid I was. It's just not worth it.
You guys may go all the way 100. Bring your hearing with you.
I could tell straight away no.2 was WAV, so clear sounding and sharp. AAC sounded bloody awful.
I would’ve bet #2 is wav (seemed to have more depth), but honestly couldn’t hear much difference between #1 and #3.
Hey Andrew! I noticed that as well, but you're right, it isn't easy to tell a difference between AAC and MP3. Thanks for watching!
SageAudio.com
Ditto to everything you said Andrew. I noticed the difference in depth as well.
I was kinda surprised that even though CZcams compresses the sound I could still tell that 2 sounded the best. 1 and 3 are pretty much indistinguishable to me.
I was right on the wav, wav had more different sounds/tones in my ears than mp3 and aac had, but I couldn't tell the difference between aac and mp3, both sounds exactly the same to me. I personally use mp3 320kbps instead of aac for my music library, I don't have a great expensive audio setup so mp3 is just fine to me. Mp3 also has a slighty punchier bass over aac (bass sounds thinner on aac or maybe it's just in my head I don't know lol) which is important to me.
While AAC is the newer more technologically advanced format the LAME MP3 encoder was worked on and perfected by fans for decades after official support ended. Long after AAC the format that was suppose to replace it was launched. Through years of refinement it came to surpass it's supposed successor as a music container format. AAC is still better at lower bitrates though which makes it better for online streaming platforms. But MP3 still has it's place for listening to music on the go.
Zonned: Well said. LAME MP3 at a high bitrate (such as vbr-0) is transparent to 99% of listeners. Yes, AAC is better than MP3 at low bitrates, but most of us don’t rip to a low bitrate, storage is so cheap that it makes sense to rip to a high bitrate that your ears will enjoy. Another bonus of LAME MP3 is portability, it will play on any device.
I like the null-test idea. But you never specified what exact AAC compression you used. there are certainly more efficient versions out there, such as the aac-lc nero.
Hey Jon R! Great point - for both the MP3 and AAC conversions I used the latest version of Logic's method. So these we're exported directly from the Logic Pro session. Thanks for watching!
SageAudio.com
@@sageaudio Did you write your own AAC encoder?
I was right, I thought track 2 was WAV.
But I got confused between track 1 and 3. Funny part is CZcams compresses the audio somehow I got track 2 right, maybe I just guessed.
This actually depends on which AAC encoder you're using, if you use iTunes apple optimization of AAC encoder it's miles better than what MP3 does. although sadly other AAC flavor optimized encoder arent even close to that.
If you take a look at something like Izotope RX to compare the original lossless VS AAC VS MP3 you'll see the bigger picture.
Even tho I wanna use Opus for my daily driver when it comes to DAP for saving space, not all supports it which is kinda sad. Opus pretty much is the replacer for AAC and MP3 in terms of saving space and sound quality
Thanks for sharing your thoughts! Yeah, definitely a lot of variables to consider
Sorry I’m commenting again but after watching your video the CZcams overlords recommended a video where this guy compared aac to mp3 but at 64kbps. Now in this test aac really shined. There was no question about which sounded better. The aac sounded decent but the mp3 was just a mess! It had that classic warble sound akin to old internet audio players from the MySpace days. So maybe this is where aac comes in handy? Just figured I’d let you guys know. Thanks again.
Hey David! No worries at all - we appreciate you leaving a comment! That's great to hear - maybe it works better at lower kbps. Thanks!
SageAudio.com
It is also not really clear which aac encoder he used. There are about 5 known ones. And trust me, the quality of those fluctuates wildly. For example, the ffmpeg encoder is barely usable even on high bitrates.
I wasn't able to tell the difference. And I'd like to believe my Bang & Olufsen has relatively high audio quality...
which one is better quality: 320kbps mp3 or 256kbps aac (itunes rip). Spectrogram shows cut-off at 20khz (Mp3 320) but shows no cut-off (aac). Can we rely on spectrogram analysis for choosing better quality ? please solve my query....
Spectrum analysis will show you what's getting cut out which is helpful! It may not sound subjectively better to someone, but it gives you an objective metric to measure. Thanks for watching!
SageAudio.com
@@sageaudio So what would be better in your opinion between them?
An mp3 encoded with LAME at vbr-0 is transparent to 99% of us. If you’re listening on earbuds, it’s a waste of space to fill up your iPhone with ALAC files or 320 mp3. I rip my CDs to lossless ALAC for archival, then convert the ALAC to mp3 vbr-0 which I load onto my iPhone. It’s a big myth that aac sound better than mp3, that is only true at low bitrates. A high-bitrate mp3 sounds every bit as good as an equal aac.
Thanks bro
I can always tell the difference between MP3 and WAV. In your example I new straight away.
Awesome thanks for watching Saul!
SageAudio.com
I guessed correctly, but only because of the previous tests in the video. I would've assumed AAC would sound superior to MP3. This is quite surprising.
Hey C. Jay! Thanks for watching. I was surprised too - maybe it was the example we used, but still strange to see that the AAC cut out more info.
SageAudio.com
Same hear. I actually knew what to listen for from hearing the null test. Pretty cool.
I preferred 2 / 3 /1 - I thought aac has better top, but mp3 sounds fuller?
Hey Little Joe! That's interesting - maybe the AAC cuts lower frequencies more whereas MP3s cut out more of the high end.
SageAudio.com
@@sageaudio It certainly sounds that way, in the example you have chosen. I have also learned that AAC, allows bit rates up to 500 kB per second?
What's the difference between AAC-LC , AAC+, AAC-ELD?
They're different algorithms that delete different parts of the signal when created the lossy format. AAC-low complexity is an early version of the AAC.
AAC+ is used by radio stations and a lot of streaming services. AAC-ELD is typically used when encoding audio for video.
SageAudio.com
Thank you for your share. If you have the need to convert between AAC, MP3, AIFF, FLAC, WAV, M4A etc, Avdshare Audio Converter can help.
Thanks for watching Shelley!
SageAudio.com
Gotta say about apple what you want, but they do care about music. Best clue is in their own aac encoder which is still the best one.
Vorbis is still inferior to it and opus is really good but not really made with archiving in mind. More for streaming.
Of course, when you listen to it it doesnt really matter, but on the spectrum it is visible. The more bitrate you give to aac, the better it reaches the lossless quality. Opus on the other hand is already really great at 96kbit/s but unfortunately that is also about the point of diminishing returns.
Thank you for watching and sharing your thoughts on this topic! I agree, Apple does seem to care a lot about music!
SageAudio.com
@@sageaudio It isnt really my opinion but these listening tests kinda validate what i said: hydrogenaud.io/index.php?topic=120166.0
You can see that while on < 128kbit/s opus is better, aac overtakes opus around 128 kbits.
I think there is a reason the official test on the opus website only looks at 96 kbit/s.
listening-test.coresv.net/results.htm
So buying the White Album on MP3 is superior than Apple's version?
Thanks for watching! It would depend on how it was encoded and what version of the AAC was being used. These were just tests from Logic Pro X and the encoder being used there, but Apple also offers a "Apple Masters Droplet" which is great for converting a file to AAC.
SageAudio.com
I could discern. I could tell 1 was better than 3 but couldn’t believe 2 was AAC because you showed otherwise.
Thanks for watching!
SageAudio.com
i knew which track was the WAV to my ears the aac file sounded better than the mp3
What is better cbr or vbr in quality?
Thanks for watching! Constant Bit Rate or CBR is better. VBR is variable - meaning to save space the encoder will delete info at various points whenever it seems like important/less perceivable to the listener.
At the blind test I literally guessed all file types first try
Nice!
Sageaudio.com
AAC vs DDP ??
please, I haven't found any on CZcams
can aac at some point beat the dolby ?
CZcams compression ruins this test anyway. My results: track 1: opus 251, track 2: opus 251, track 3: opus 251.
How can this happen? I checked the internet and always told that AAC (.M4A) is always better than MP3, all of them said 128kbs .M4A is better than 160kbs .mp3! but I can clear hear your sample that 320kbs.mp3 is better than 320kbs.m4a!
Awesome! Thanks for watching!
SageAudio.com
This is because the encoder is also a factor in this. I think it was mentioned in other comments as well that Apple's qaac encoder is considered to be the best quality wise.
Could you please repeat this experiment by comparing a QAAC encoded AAC, @SageAudio?
I clearly hear the mp3 even tho the youtube compression. Noice.
Hey Mórocz! Thanks for watching - cool to hear that despite CZcams's compression you could still tell a difference - I was wondering how that would affect this video.
SageAudio.com
The middle one was the odd one out so it had to be the wav.
My first thought was AAC, WAV, the MP3. But then I second guessed myself and though AAC, MP3, WAV.
Test done with Sennheiser HD8's.
Nice! Thanks for watching
I though it was
1-AAC
2-Joint Stereo MP3
3-WAV
😨 -- CZcams compression cannot stand this
Hey Tuggaboy! Thanks for watching - it was hard to tell which was which!
SageAudio.com
If i were to swap my flacs for lossy files, i would go with the aac encoder from apple.
I guessed all 3 correctly listening on my phone without headphones lol
Nice!
SageAudio.com
I would‘ve said 1 is MP3, 2 WAV and 3 AAC. I thought the 3rd Track was sounding the worst, that‘s why I went with AAC.
Nonetheless, even with the CZcams compression, the WAV stands out IMO. I watched this entirely on my iPhone 12 Pro Max with no earphones.
Thanks for watching!
SageAudio.com
#1 deinately sounded better than #3 for me, seems even though they recommended mp3, to my ears aac sounded better on the blind test.
But all the examples are the some compression when you listen on youtube. You can't judge based on the video. You would need to get the raw files and download them.
Theres no need to use any lossy format, this isnt 1995
My tip: 1.MP3, 2 AAC, 3WAV
Hey Marek! Awesome - thanks for leaving a comment!
SageAudio.com
Best stereo audio codec is opus.... Aac is hyped for apple users while opus is way much better and open source
Thanks for watching!
Opus for me
Thanks for watching James!
SageAudio.com
Vorbis 500kbps
So many things were wrong with the test, including that lossy re-encoding on CZcams is inevitable (now primarily to Opus). But also that virtually nobody uses 320 kbps, cancelling out the waveform to reveal what remains isn't a direct indicator of perceived quality (what lossy codecs aim for), the tiny boost in dB is intentional to counteract the dampenning results of encoding and will never be done to a point of audible clipping, and so on. Exhaustive testing has been done and AAC is clearly superior. It takes 128 kbps Nero AAC (the best) to match 192 kbps LAME MP3 (the best), and these score as high as 320 kbps versions >99% of the time so there's little reason to use higher bitrates, let alone 320 kbps.
According to hydrogenaudio, the apple aac encoder is still the best one. I dont really use lossy files because i dont want to commit to one of them. But yeah that encoder seems to be a valid challenger if i were to swap my lossless collection. Wav is too space - inefficient and unnecessary for me. I mean, you can literally CALCULATE what the filesize is going to be with the lengths, channel number, sampling rate, and bit depth no matter the actual contents. Flac/alac seems to be a nobrainer to me.
Opus is so good but at the same time also so bad. It sounds awesome at low bitrates but at the same time no one should use it to archive their music.It just doesnt scale well to high bitrates. Also they force you to resample to 48 khz which is great if you want STREAMING. But for archiving, thats a no go and really shows the priorities of the devs.
@@kartoffelbrei8090 I can't tell 128 aac or 192 mp3 apart from the source 99% of the time so I use them, but according to double blind testing with audiophiles on high end equipment 128 kbps AAC can be reliably distinguished from the CD source so they should use flac since storage space and bandwidth is no longer an issue. But I can detect the difference between Apple 128 vs Apple 256 or flac, but not nero 128. It's likely because the high pass filter is implemented better by Nero than Apple and LAME MP3, both of which look much sharper on graphs.
@@brandon1902 I wouldnt really call this high pass filtering i think that would be too generous.
I still think there is good reason to go lossless. You dont really know what the codec acutally does with CELT and psychoacoustic models whereas lossless guarantees no losses (duh).
Yeah i know, its really inefficient compared to lossy, but i dont want to find out that the codec generated some kind of artifact in my music some day later.
I had that happen before with ffmpeg AAC.
Thank you all so much for watching this video and for commenting! Also, thanks for starting a dialogue about this topic, there are a lot of ideas to explore and consider.
SageAudio.com
wav wasn't as good as the other two formats
Thanks for watching!
SageAudio.com
I think the AAC sounded better than the other two. And the wave and mp3 sounded exactly alike
Thanks for sharing!