Okay, so I have some original flacs ripped from cds. I took an abx test and turns out I can't see any difference. But still, I want a little safeguard just in case. So I experimented with some codecs, notably opus and vorbis, and I noticed something. Whenever I use the highest quality available,(vorbis q10 ~500kbps and opus 512kbps), and scan the files with Spek, vorbis has cut off a lot less of the high end frequencies, while opus cuts it off at 20khz, just like an mp3 320kbps. Why is that? Wasn't opus supposed to be the evolution of vorbis? So why is it outperformed by vorbis?
Also, just another question, does transcoding from vorbis degrade considerably audio quality?
Using latest libraries on soundkonverter, Linux.
Thanks so much! Best wishes!
No one can hear frequencies above 20kHz. Why should Opus waste space encoding frequencies that no one will be able to hear?
If you want a pretty spectrogram, encode your audio with an image codec like JPEG instead. It won't sound very good, but it will look pretty.
Or stick to lossless.
I can't see any difference.
You mean you can't
hear a difference? ;)
Lossy compression has to throw-away
some information. If you keep high frequencies that you can't hear you might end-up throwing-away something more important.
Perceptual encoding mostly works by throwing-away sounds that are masked (drowned-out) by other sounds. As I understand it, at most settings MP3 takes a shortcut and assumes the highest frequencies will be masked and it throws-away the highest frequencies with little or no analysis.
If you hear a compression artifact or if you can hear a difference in an ABX test, it's usually NOT the loss of high frequencies that you hear (assuming a reasonable quality/bitrate setting) but it's the easiest thing to
measure. Looking a the spectrum is NOT not a good way of comparing sound quality. It's probably easy to make a bad-sounding compression algorithm that makes nice-looking spectrums.
No one can hear frequencies above 20kHz. Why should Opus waste space encoding frequencies that no one will be able to hear?
I can hear frequencies above 20KHz, confirmed by ABX, with and without 20KHz lowpass. So it does matter to me to keep them when full quality matters. Opus is highly efficient, and is a lot like a streaming successor to MP3, but Vorbis aoTuV at higher bit rates sounds clearer to my ears than most other formats. Not everyone will have the same opinions on which codecs are more transparent to them personally but Vorbis will likely get the job done between -q8 and -q9.5 and reasonably keep some of the higher frequencies.
I can hear frequencies above 20KHz, confirmed by ABX, with and without 20KHz lowpass.
How can you hear frequencies above 20 Khz if a filter is filtering everything above 20 Khz? I'm confused.
No one can hear frequencies above 20kHz.
That is inaccurate - test subjects have heard 28 kHz - but in this study it took more than 90 dB: https://asa.scitation.org/doi/pdf/10.1121/1.2761883
And yes it is a well-reputed lab.
Thanks everyone for your feedbacks! What codecs do yall use and at what bit rate?
I can hear frequencies above 20KHz, confirmed by ABX, with and without 20KHz lowpass.
This guy is smoking some good shit, lol.
Anyway, nobody is using Vorbis anymore. It is dead format.
If you want compatiblity and good all-around format stick with MP3.
Use latest LAME encoder and use V2 preset. You will get ~192 kbit/s files that work everywhere and sound great.
If your equipment has been made in the past ~15 years you can use AAC-LC.
Recommended encoder is Apple (qaac). Ideal bitrate is around 144-160 kbit/s.
Third "relevant" format acording to HydrogenAudio is Opus. This is totally false in real life.
Even today, many programs have problems with playing Opus files. It is really only useful for VoIP applications.
I can guarantee you that every single person on HA that is using Opus is very familiar with compatibility
problem but they also have lossless archive so they can encode to any format that they want.
Main point of lossy files is transparency at smallest bitrate. Lossy codecs achieve this at 192 kbit/s.
Everything about 192 kbit/s kills the point of lossy files.
Thanks everyone for your feedbacks! What codecs do yall use and at what bit rate?
Lossless for to keep - transcoding to two-figure bitrate for on-the-road. Opus because it impressed me ...
and because artists don't use it much for free downloads, so by using Opus I "know it is my own transcode", and those files are disposable.
Anyway, nobody is using Vorbis anymore. It is dead format.
As long as the biggest streaming platform is using Ogg/Vorbis I don't consider it a dead format...
Personally I still use it for lossy because it has bettter gapless playback support than mp3 and Opus, and because it has better compatibility than Opus.
Anyway, nobody is using Vorbis anymore. It is dead format.
Personally I still use it for lossy because it has bettter gapless playback support than mp3 and Opus, and because it has better compatibility than Opus.
Same for me.
And it's far from being dead especially with Android running on ~80% of the mobile devices.
But I agree that MP3 is the best choice for best compatibility.
At high bitrates Vorbis is indeed better than Opus which was quite a revelation for me when I actually tested both.
At high bitrates Vorbis is indeed better than Opus which was quite a revelation for me when I actually tested both.
I'm sure you have some logs to offer.
At high bitrates Vorbis is indeed better than Opus which was quite a revelation for me when I actually tested both.
I'm sure you have some logs to offer.
I don't have any tools akin to SSIM for videos but I trust spectrograms and auCDtect output which puts Vorbis above Opus.
At high bitrates Vorbis is indeed better than Opus which was quite a revelation for me when I actually tested both.
I'm sure you have some logs to offer.
I don't have any tools akin to SSIM for videos but I trust spectrograms and auCDtect output which puts Vorbis above Opus.
How many times have you been pointed at this forum's terms of service (https://hydrogenaud.io/index.php?topic=3974)?
You have not "actually tested both" for being any "better".
How many times have you been pointed at this forum's terms of service (https://hydrogenaud.io/index.php?topic=3974)?
You have not "actually tested both" for being any "better".
I have to agree with you then and withdraw my outrageously wrong statements.
How many times have you been pointed at this forum's terms of service (https://hydrogenaud.io/index.php?topic=3974)?
You have not "actually tested both" for being any "better".
I have to agree with you then and withdraw my outrageously wrong statements.
It's kind of sad for us forum lurkers to see yet another one succumb to confirmation bias instead of performing some really basic scientifically valid audio comparison testing.
Its not entirely wrong. From the perspective of very high bitrate one may prioritize spectrum over efficiency.
Extra psy bells and whistles may be perceived as a liability . In the most tuned coders- mp3 lame, vorbis, mpc ; the higher the quality level the less psychoacoustics are used .
It's not outrageously wrong, it just needs ABX to back it up.
In blind comparison, I can hear aoTuV Vorbis at 268+ kbps tend to deliver clearer sound to the original than the best Opus or LC-AAC encoders at the same rate, whether it be a matter of stereo fidelity or transient fidelity, or that 20KHz cutoff that I can hear when played in a car (using the AUX jack not Bluetooth). There hasn't been a listening study at that bit rate, so it might just be me and my tastes.
I can hear frequencies above 20KHz, confirmed by ABX, with and without 20KHz lowpass.
This guy is smoking some good shit, lol.
Scientific studies on human hearing found that frequencies above 20KHz are difficult to hear, not that you never hear them. The maximum frequency response in air has been measured from 12Hz at the lowest recorded, to 28KHz in laboratory conditions, according to The Journal of the Acoustical Society of America (https://asa.scitation.org/doi/10.1121/1.2761883). (Of course, they're talking very high SPL.) The important distinction is that high and low frequencies are easily
masked, not necessarily inaudible. With most listeners in the test able to hear 22KHz, I know I'm not alone at being able to ABX a 20KHz cutoff.
Lots of claims - not a single ABX. Something is very wrong here.
People like you are the reason why we have stuff like MQA and 24-bit/192kHz downloads...
It's not outrageously wrong, it just needs ABX to back it up.
In blind comparison, I can hear aoTuV Vorbis at 268+ kbps tend to deliver clearer sound to the original than the best Opus or LC-AAC encoders at the same rate, whether it be a matter of stereo fidelity or transient fidelity, or that 20KHz cutoff that I can hear when played in a car (using the AUX jack not Bluetooth). There hasn't been a listening study at that bit rate, so it might just be me and my tastes.
I think you can probably share the methodology and results so that Mr Birdie can perhaps get started on HIS double blind comparison.
Its not entirely wrong. From the perspective of very high bitrate one may prioritize spectrum over efficiency.
Extra psy bells and whistles may be perceived as a liability . In the most tuned coders- mp3 lame, vorbis, mpc ; the higher the quality level the less psychoacoustics are used .
Will be nice to have some actual data on how these codecs work rather than intellectual theories.
People like you are the reason why we have stuff like MQA and 24-bit/192kHz downloads...
Those MQA believers/shills actually don't care about >20kHz performance.
https://www.audiosciencereview.com/forum/index.php?threads/mqa-deep-dive-i-published-music-on-tidal-to-test-mqa.22549/post-808088
They worship the so-called time-domain performance, and think audio files don't follow the "rule" of triangular spectral decaying trend are not music.
https://www.audiosciencereview.com/forum/index.php?threads/mqa-deep-dive-i-published-music-on-tidal-to-test-mqa.22549/post-802823
https://www.audiosciencereview.com/forum/index.php?threads/mqa-deep-dive-i-published-music-on-tidal-to-test-mqa.22549/post-803661
I don't know I've been listening to so many non-music for so many years, big thanks to those MQA shills teaching me what is music! LMAO.
https://www.audiosciencereview.com/forum/index.php?threads/mqa-deep-dive-i-published-music-on-tidal-to-test-mqa.22549/post-802352
https://www.audiosciencereview.com/forum/index.php?threads/mqa-deep-dive-i-published-music-on-tidal-to-test-mqa.22549/post-807177
Of course, after losing the debate and thread-banned mansr, Amir locked the thread again, and continue to publish MQA DACs reviews with nice looking APx555 measurements.
Also a recommended thread for spectrophiles. Three pages to expose a liar.
https://hydrogenaud.io/index.php?topic=111736.0
Its not entirely wrong. From the perspective of very high bitrate one may prioritize spectrum over efficiency.
Extra psy bells and whistles may be perceived as a liability . In the most tuned coders- mp3 lame, vorbis, mpc ; the higher the quality level the less psychoacoustics are used .
Will be nice to have some actual data on how these codecs work rather than intellectual theories.
You're too polite to called it "intellectual theories", I call them what they are.
BS.
All previously years they were crapposting all over the place (without any proof) how Opus or any other encoder can't compete with Musepack or LossyWAV , "king of transparency" at high bitrates. "Opus/AAC/(put your codec here) are for low bitrates ... bells and whistles .. b$$ b$$"
So I decided to perform actually test for this kind of
people (you know, theorists ... $**t posters) and break every encoder in all possible ways with hard samples that I only could.
Ironically Opus did transparent in all of samples. :P hahahahaha And
a king of transparency Musepack was on par with AAC/Opus/Vorbis/xHE-AAC
Personal blind listening test – MultiCodec at ~192 VBR kbps (https://hydrogenaud.io/index.php?topic=120007.0)
If anybody has a better proof that this, go on and post your tests ... man I'm begging you to post your tests ...... but please keep your "perespectives and philosofical question of life" to yourself. Because nobody gives a finger about them.
Hey Igor, do you know of any tests with 256k? I really wanted to know how codecs perform at that bit rate, cause I always used at least 256k. Thanks for your participation so far!
No, I don't know.
You're too polite to called it "intellectual theories", I call them what they are. BS.
I beg to differ. It is hardly controversial to say that lossy codecs have improved over thirty years (like, Monty's classic-at-this-forum, search for [2]
and read the footnote: https://web.archive.org/web/20200417180912/https://people.xiph.org/~xiphmont/demo/neil-young.html )
... so what did they start lossy codecs at? Some pretty good ideas of what should work, and some pretty good ideas of what could
improve. What actually improves, well you can test - searching for improvements (better than blind search), that is brainwork.
And if you have found an improvement up to bitrate B, but have not yet picked up a B+100 troublesome signal, what then? Inferring how to extrapolate from data is harder than your statistics 101, but it surely isn't BS.
In the above link, Monty also quotes some anonymous user:
"[The Sampling Theorem] hasn't been invented to explain how digital audio works, it's the other way around. Digital Audio was invented from the theorem, if you don't believe the theorem then you can't believe in digital audio either!!"True or not ... that puts it on intellectual theory. The sampling theorem does not tell you
how high limiting frequency is adequate for musical reproduction for the human ear (with all the filters that were invoked in the technology's infancy), that is an empirical exercise.
Well, I conducted various ABX over the last few days and a hearing test today, and the blind test results weren't passable (11/16 at best, p-value >10%). Easy ear fatigue. I could blame the fact I'd moved to a new house (major differences in resonance) and the window air conditioners, but I'm going to conclude that at age 37 I can't hear ultrasonic frequencies well enough to ABX the tracks I tested and a brick wall 20KHz lowpass. Because no formal tests were done 2 years ago, I retract my claim on that as a statement of fact.
In the hearing test, I could hear a pulsating 20.5KHz sine, using Audacity sampling at 96KHz. The boost required was so high that going higher could damage my ears, but I can safely conclude that I can hear above 20K, just barely. The dropoff after 19KHz was very steep.
I never claimed ultra high res was necessary; in fact the test in the article I linked made it clear that not one person could hear 30KHz. That would mean for even the most ideal conditions a maximum master sample rate of 60KHz. There's no reason for 192KHz masters unless you want to entertain animals. And the thing about controlled environments with extremely high Signal Pressure Levels is that they don't reflect any kind of normal listening experience. The kind of levels required for 24KHz are dramatic enough that I'll probably never hear of anyone who can tell the difference between a 96KHz FLAC and 48KHz of the same source. Yes, 48KHz is good enough. (I just wish Opus gave the option to encode more bands.)
As for MQA, I think Mnyb on ASR described it best: "a closed source licensing racket that no one needs."
20.5KHz sine [...] I can safely conclude that I can hear above 20K, just barely.
Barely hear and barely above: it is only a quarter tone.
Going from there to saying what sample rate is adequate, is a question of era and technology. Nowadays we can record at high sample rate and then use a steeeeeeep filter in the digital domain to eliminate aliasing before resampling to end-user format.
But if you had to use an analogue lowpass between your microphone and your 44.1 kHz ADC, it would have to kick in much lower. So by all means,
record at higher sample rate.
In the hearing test, I could hear a pulsating 20.5KHz sine, using Audacity sampling at 96KHz. The boost required was so high that going higher could damage my ears, but I can safely conclude that I can hear above 20K, just barely. The dropoff after 19KHz was very steep.
I'd like to emphasize that just because you can hear 20.5 kHz in ideal conditions, it doesn't mean it's not completely masked in actual music.
There are tests in this forum putting high bitrate Opus above Vorbis in transparency.
20 kHz low pass filter on Opus is more than reasonable. Musical instruments do not purposefully emit above 20 kHz. Human voice does not reach that frequency either. >99% humans do not notice / cannot notice >20 kHz harmonics. >99% speakers are specced within the range below 20 kHz.
The fact that you might be able to hear isolated >20 kHz tones at high volume under silent conditions bears no resemblance whatsoever to whether you can notice >20 kHz harmonic tones while enjoying music. Try hearing that tone at low volume while enjoying a track.
Regardless, if you want to raise or remove the Opus low pass filter, you can enter custom parameters when encoding. If your only reason for using Vorbis is that it doesn't use a low pass filter, you can use Opus with custom parameter instead.
I use both Opus at (nominal) 192 and Vorbis at (nominal) 256, depending on the album. The reason I use both is simple, Opus has a glitch in tracks that do not transition with silence. Does not have perfect gapless.
Third "relevant" format acording to HydrogenAudio is Opus. This is totally false in real life.
Even today, many programs have problems with playing Opus files. It is really only useful for VoIP applications.
Android cell phones have seamless integration of both Vorbis and Opus files in current year.
I can hear frequencies above 20KHz, confirmed by ABX, with and without 20KHz lowpass.
This guy is smoking some good shit, lol.
Anyway, nobody is using Vorbis anymore. It is dead format.
If you want compatiblity and good all-around format stick with MP3.
Use latest LAME encoder and use V2 preset. You will get ~192 kbit/s files that work everywhere and sound great.
If your equipment has been made in the past ~15 years you can use AAC-LC.
Recommended encoder is Apple (qaac). Ideal bitrate is around 144-160 kbit/s.
Third "relevant" format acording to HydrogenAudio is Opus. This is totally false in real life.
Even today, many programs have problems with playing Opus files. It is really only useful for VoIP applications.
I can guarantee you that every single person on HA that is using Opus is very familiar with compatibility
problem but they also have lossless archive so they can encode to any format that they want.
Main point of lossy files is transparency at smallest bitrate. Lossy codecs achieve this at 192 kbit/s.
Everything about 192 kbit/s kills the point of lossy files.
I'd rather use 192kbit/s Musepack and with Android being so popular compatibility not a issue these days, Since MPC seems more robust on some Noise/transient samples that Opus groans on. But overall Vorbis is worse than AAC/MP3 by long shot not even transparent on most samples at 256kbps while AAC/MP3 are to be blunt.
I'd rather use 192kbit/s Musepack and with Android being so popular compatibility not a issue these days, Since MPC seems more robust on some Noise/transient samples that Opus groans on. But overall Vorbis is worse than AAC/MP3 by long shot not even transparent on most samples at 256kbps while AAC/MP3 are to be blunt.
@ani_Jackal3: You are one of those guys I usually skip as you come around and like a high priest you trot out these statements as if you were some guy who did a lot of tests. Just that you do this now everywhere on this forum, and this is the point I have enough.
Prove me wrong and show us your tests and proper analysis on these claims. Show us those "most samples" where 256 kbps Vorbis files are not transparent with ABX logs. Also show us your Musepack vs Opus tests wher Opus "groans", but Musepack shines.
If you cannot refer to those tests, I would prefer if you could just stop spreading BS.
Interesting. Vorbis went through several tunings. Garf's GT3 for Q5 -10, Then AOtuv. By that time
I recall a post by pio2001 saying he knows badvilbel as Q6 killer (i confirmed it) but none for Q7 +
The old testers were pretty hardcore.
But overall Vorbis is worse than MP3
Now I've truly seen everything.
But overall Vorbis is worse than AAC/MP3 by long shot not even transparent on most samples at 256kbps while AAC/MP3 are to be blunt.
Very interesting, I've done a few personal tests(not posted on HA) in the past and I never found an example where MP3 outperformed Vorbis
I invite everybody report ani_Jackal3's posts for violation of TOS 8 (https://hydrogenaud.io/index.php?topic=3974#post_tos8)
I invite everybody report ani_Jackal3's posts for violation of TOS 8 (https://hydrogenaud.io/index.php?topic=3974#post_tos8)
Don't do it. Have an open mind. It would be very interesting to hear back / get some samples etc.
Sure, just hold your breath for another year. "Be an open mind" Cmon
We had enough of chit-chat with him and similar, any dialog is futile at this point.
I mean, let's be honest. Vorbis worse than mp3 is a preposterous statement. Nutcase level.
I mean, let's be honest. Vorbis worse than mp3 is a preposterous statement. Nutcase level.
Back in the day around 2003 or so. it was proposed that vorbis is not mature compared to
mp3 - particularly preset standard / extreme (V2 etc). Many stuck with mp3 and even mpc.
There was a HF noise until Q5 or 6. Since then it has been tuned so these things are fixed or
reduced. Could be that edge cases exist today.
Back in the day around 2003 or so. it was proposed that vorbis is not mature compared to
mp3
Back in 1995 or so, it was proposed that MP3 was not mature compared to MP2.
So what? "Could be that edge cases exist today"
Your chit-chat isn't any better than Jackal3's systematic violation of TOS8 during years.
Cut the B chit-chat S, sir.
You mention 2003 but you don't mention 2020's Personal blind listening test – MultiCodec at ~192 VBR kbps (https://hydrogenaud.io/index.php?topic=120007.0) where Vorbis was clearly superior to MP3.
How convenient from your part, sir.
Theres no tos 8 whatsoever. I stated nothing but facts without going into every detail.
One listening test doesn't represent all scenarios. So there can exist an edge case where
a superior codec does worse than inferior one etc .
Theres no tos 8 whatsoever. I stated nothing but facts without going into every detail.
One listening test doesn't represent all scenarios. So there can exist an edge case where
a superior codec does worse than inferior one etc .
It wasn't claimed that Vorbis is worse than MP3 in
edge cases, it was claimed that:
But overall Vorbis is worse than MP3
(emphasis mine)
And "overall" requires a verification with a listening test, not just speculation about "edge cases".
I am with IgorC on this. You only add to the discussion if you back up that claim as it goes against all documented tests in the last 5+ years. There could be cases, but at he same time there could NOT be such cases. The onus is on you to prove that there are such cases.
Even one case where MP3 shines and Vorbis needs proof. Again, saying that there could be such cases says nothing as by the same token there could not be such cases. These are weighless statements.
I can back it up. For vorbis ,Bavilbel is worse than mp3 until Q7. So there are cases.
That is why I was interested with ani_jackal [big] claims as it goes against the
accepted narrative.
It would be nice to have a link to a post with the ABX and the sample. This is not very specific.
But I have this killer sample. So if I have time I might try Vorbis latest Aotuv at Q4 and which Lame setting? (Q4 as I assume I will hear the diff there.)
It would be nice to have a link to a post with the ABX and the sample. This is not very specific.
But I have this killer sample. So if I have time I might try Vorbis latest Aotuv at Q4 and which Lame setting? (Q4 as I assume I will hear the diff there.)
I don't have it. It was bad enough that abx wasn't needed (on the first noisy part). I remember it that clear.
vorbis Q4 - 6 vs similar rates mp3. However at Q7 it was solved.
I can back it up...
...2 days later...
I don't have it...
Well, it was a pleasure to talk to You.
Bye
I can back it up...
...2 days later...
I don't have it...
Well, it was a pleasure to talk to You.
Bye
When I get to it I will post the links. it is confirmed by others and AOTUV
developer. You are not holding a stopwatch on me.
And others know me. I am reliable. Look at the angels falls first sample, the berlin drug sample. They
are mine. The efforts to tune wavpack lossy by finding samples.
My time, my efforts when I could have done other things.
So lose your arrogance a bit.
It's not a first time You make unverifiable comments. Actually, there are many of them.
So stop f talkin and start actually backup your statements.
I am with IgorC on this. You only add to the discussion if you back up that claim ...
Thank You! Most of us here agree on that.
Yeah and when I do what will you do then ?
It's not a first time You make unverifiable comments. Actually, there are many of them.
So stop f talkin and start actually backup your statements.
I am with IgorC on this. You only add to the discussion if you back up that claim ...
Thank You! Most of us here agree on that.
No, Most DO NOT agree. That is a lie . People that know me since 2004 , know I am reliable.
I been here 16 years + . You don't come and tell me how its done .
No, Most DO NOT agree. That is a lie . People that know me since 2004 , know I am reliable.
I been here 16 years + . You don't come and tell me how its done .
Huhauhauhauhauhauha
:D ;D :))
Man, just see what people have responded to You in this topic.
https://hydrogenaud.io/index.php?topic=60956.msg745595#msg745595
https://hydrogenaud.io/index.php?topic=53335.msg478296#msg478296
https://hydrogenaud.io/index.php?topic=56380.msg506486#msg506486
https://hydrogenaud.io/index.php?topic=6023.msg65211#msg65211
I will test your test case as per above. So I will play ball. but none of your links has any ABX proof that they can actually ABX the difference even on that sample. Yeah, I hear it is easy. Then give us some proof.
It seems like a bunch of you are in an echo chamber. This is not how Science works. Just parrotting the same thing along with a bunch of others (who might have picked up this from you) proves nothing. This is not a competition how many others I can make to parrot the same thing.
The thing is Vorbis is better than Lame is 99% by a large margin. You will have a tough going to prove otherwise. So even if there is one sample, at least we should keep in perspective (and have proper documentation of that rare case).
Yes vorbis is better, I never said otherwise. It was another poster.
Since aotuv, tests show its very good. That's what I said.
i said 'may be edge cases' like this that I stumbled on in the past.
Why should you not believe me ?
I find an issue with software ; lossless or lossy. Say wavpack. Okay ?
I contact the developer - a very respectable individual BTW.
I don't get asked to prove anything. He will check it to confirm because there's an
interest to make software better.
I contact the developer - a very respectable individual BTW.
I don't get asked to prove anything.
That's how that one particular developer manages things. That's ok for him.
You are not contacting him right now, You are posting in Hydrogenaudio and You should be TOS8-compliant > Terms Of Service (https://hydrogenaud.io/index.php?topic=3974)
Even if You were Paul McCartney ( he is even more popular than You) you would still need to provide a
downloadable sample and ABX log. TOS8 isn't optional.
P.S. We don't have this discussion with Kamedo2, Guru, /mnt even me :P .... even if they are reliable they still provide all kind of information. No need to ask anything... everything is available!
Yes to prove one codec is better than another codec. I never claimed it.
I said codec vorbis has bug on this sample . Like amnesia & fsol samples mpc q5 was
worse than lame V2 but no one said mp3 is better than mpc ..
BS; LancerMod(SSE2) (based on aoTuV [20110424][1.3.5])
Q4 - pretty easy. Bitrate 154k
foo_abx 1.3.4 report
foobar2000 v1.5.7
2021/07/09 01:31:39
File A: G:\Hdd-2\Music\abx tests\badvilbel.flac
File B: G:\Hdd-2\Music\abx tests\badvilbel.ogg
01:31:39 : Test started.
01:31:51 : 01/01 50.0%
01:31:58 : 02/02 25.0%
01:32:00 : 03/03 12.5%
01:32:07 : 04/04 6.3%
01:32:10 : 05/05 3.1%
01:32:12 : 06/06 1.6%
01:32:15 : 07/07 0.8%
01:32:19 : 08/08 0.4%
01:32:23 : Test finished.
----------
Total: 8/8 (0.4%)
Q6 - I started here and it seemed harder than I remember. Vorbis got better since 07 or I got worse (probably)
So.. after going to Q4 , I decided to do another round of Q6
Q6 -first test. Bitrate 252k
foo_abx 1.3.4 report
foobar2000 v1.5.7
2021/07/09 01:26:47
File A: G:\Hdd-2\Music\abx tests\badvilbel.flac
File B: G:\Hdd-2\Music\abx tests\badvilbel.ogg
01:26:47 : Test started.
01:26:57 : 00/01 100.0%
01:27:03 : 01/02 75.0%
01:27:43 : Trial reset.
01:28:06 : 01/01 50.0%
01:28:16 : 02/02 25.0%
01:28:36 : 03/03 12.5%
01:29:03 : 04/04 6.3%
01:29:22 : 05/05 3.1%
01:30:16 : 05/06 10.9%
01:30:29 : 06/07 6.3%
01:30:48 : 07/08 3.5%
01:31:04 : Test finished.
----------
Total: 8/10 (5.5%)
Q6 - part 2, It became clearer.
foo_abx 1.3.4 report
foobar2000 v1.5.7
2021/07/09 01:33:45
File A: G:\Hdd-2\Music\abx tests\badvilbel.flac
File B: G:\Hdd-2\Music\abx tests\badvilbel.ogg
01:33:45 : Test started.
01:34:05 : 00/01 100.0%
01:34:19 : 01/02 75.0%
01:34:30 : 02/03 50.0%
01:34:37 : 03/04 31.3%
01:34:52 : 04/05 18.8%
01:35:17 : 05/06 10.9%
01:35:34 : 06/07 6.3%
01:35:54 : 07/08 3.5%
01:35:56 : Test finished.
----------
Total: 7/8 (3.5%)
15 / 18 for Q6
trial reset was I decided to use a different position for playback.
For that position score is 7/8 + 7/8 = 14/16
total all: 8/10 + 8/8 + 7/8 = 23/26
Quality 5 bitrate 204k, fairly easy I am warmed up
foo_abx 1.3.4 report
foobar2000 v1.5.7
2021/07/09 02:09:37
File A: G:\Hdd-2\Music\abx tests\badvilbel.flac
File B: G:\Hdd-2\Music\abx tests\badvilbel.ogg
02:09:37 : Test started.
02:09:46 : 01/01 50.0%
02:09:50 : 02/02 25.0%
02:09:54 : 03/03 12.5%
02:10:01 : 04/04 6.3%
02:10:05 : 05/05 3.1%
02:10:09 : 06/06 1.6%
02:10:18 : 07/07 0.8%
02:10:30 : 08/08 0.4%
02:10:33 : Test finished.
----------
Total: 8/8 (0.4%)
ani_Jackal3 has just disappeared now that his unfounded statements were laid bare. It is good for the forum that we have less BS.
both not worth attention
I just want to put my two cents in on this...
As a test, I encoded the same audio file into four codecs, MP3, Vorbis, Opus, and AAC, at 128kbps and 64kbps. Relaying the facts based off of a spectrogram:
* 64kbps Vorbis filtered out frequencies past around 15kHz
* 64kbps MP3 filtered out past 11kHz and 128kbps MP3 filtered out past 16kHz
* Vorbis 128kbps, AAC, and Opus didn't filter out much
* There is a part where the right channel is the inverse of the left channel -- Opus stays the same in quality, AAC filters it out entirely, and Vorbis and MP3 drop greatly in quality.
I personally dislike MP3, because at almost any bitrate, it brickwalls a lot of frequencies. At 128kbps, you really won't be getting much past 16kHz in any audio file, and less if you chose to use simple stereo instead of joint. I can see this in these spectrograms, and hear it too with the 64kbps MP3.
From my personal experience, Opus does the best at low bitrates (64kbps here), almost tied with AAC SBR+PS. However, it's up to you to determine which sounds best to you. Once, I encoded a track with something like 48kbps Vorbis and there was a lot of high frequency content not present in the original song. It actually made the track sound better to me, I shared it with someone else and they didn't like that. It's entirely subjective.
Thx for contributing another baseless assessment on this thread which is rife with this. There is no point looking at the spectogram. The fact you look at the spectogram means that you do not bother to do blind tests, and you learnt nothing on this forum.
Determining which codec is generally better, based on personal experience only, can lead to leveling out the differences between them after conducting a listening test in a nursing home. A burger from McDonald's can satisfy your hunger, just like a home-cooked meal, but the technologies (https://www.mcdonalds.com/gb/en-gb/help/faq/do-any-of-your-products-contain-trans-fats.html) used to facilitate mass production can be harmful in the long run, which neither your tongue nor your stomach will tell you right away. Lossy encoding is about making an approximate copy of the original that is convincing enough, but satisfaction does not mean that everyone stops being interested in what they have lost. Unfortunately, such interest is more often met here with ridicule or a superficial reply like “inaudible therefore useless”, which distinguishes Hydrogen from, for example, Gearspace (https://gearspace.com/board/so-much-gear-so-little-time/12423-does-frequency-response-above-20-khz-really-matter.html). Try cutting off the high frequencies (https://www.tnt-audio.com/casse/life_above_20khz.pdf) that you don't need sharply, say above 16 kHz (obviously, using Butterworth filter with a slope of 72 dB/octave), and you'll find that the remaining frequencies that you hear are affected in a questionable way. Some people tolerate and even enjoy porn, which is also an approximation (of intimate relationships), but the possible consequences (https://www.google.com/search?q=pornography+distorts+negative+effects) can harm your well-being, so consider spreading the awareness instead of “it's up to your skin flute ears”.
A burger from McDonald's can satisfy your hunger, just like a home-cooked meal, but the technologies (https://www.mcdonalds.com/gb/en-gb/help/faq/do-any-of-your-products-contain-trans-fats.html) used to facilitate mass production can be harmful in the long run, which neither your tongue nor your stomach will tell you right away. Lossy encoding is about making an approximate copy of the original that is convincing enough, but satisfaction does not mean that everyone stops being interested in what they have lost. Unfortunately, such interest is more often met here with ridicule or a superficial reply like “inaudible therefore useless”, which distinguishes Hydrogen from, for example, Gearspace (https://gearspace.com/board/so-much-gear-so-little-time/12423-does-frequency-response-above-20-khz-really-matter.html).
The effects of a suboptimal meal (which you're using as an analogy for music consumption) can be measured throughout the day with sensors, blood samples, glycemic index, cholesterol...
Feel free to take as long as you want in the process of ABXing lossless vs. lossy. Measure heartbeat. Physiological response. If you can correlate those changes in bodily response to the actual consumption of lossless vs. lossy audio, it'll make one hell of an interesting read. Hydrogenaudio users will be stoked!