Skip to main content ↑ ↓
Topic: Looking at Opus for MP3 replacement and have questions (Read 10668 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Re: Looking at Opus for MP3 replacement and have questions

Reply #25 –
Quote from: halb27 on
b) for bitrates of say 128 kbps or more AAC is the most promising codec
It would be helpful to name reasons why You think so. 
AAC is a good choice in my opinion but Opus is too.

Re: Looking at Opus for MP3 replacement and have questions

Reply #26 –
And ogg ?

For bitrate around 128 or 160 kps, i choose ogg.
With higher bitrate like 256, i still prefer ogg over mp3.
The cutoff frequencies ae too low with mp3 (lame 3.995o Q0.5 - 256 kps : cuttoff around 18 kps))
I'am very  sensitive with high frequencies (18-19 kps)

I don't like the resampling with opus:
i have a dac with asynchronous usb (Cambridge Audio Dac Magic 100), and 98 % of my music is 44khz....
So no reason to resampling to 48 khz.... (lost of quality...)

Sorry for my bad english :S

Re: Looking at Opus for MP3 replacement and have questions

Reply #27 –
@IgorC:
Currently we don't know much about Opus' quality at higher bitrates though it won't be bad certainly.
Compare this with AAC's established quality at these bitrates. Add the wide support for AAC,  resp.  consider the fact that Opus still has a bit of a development status.

Sure with higher bitrates there are more codecs to choose from. Qualitywise it doesn't matter much which one is chosen. For universal usage on the playback side AAC and mp3 are most attractive.

@david-lisb:
Your wishes for a higher cutoff frequency can easily be fulfilled with Lame and lame3995o by using the - - lowpass option.
But if you prefer Ogg Vorbis: Go ahead.
lame3995o -Q1.7
opus --bitrate 140

Re: Looking at Opus for MP3 replacement and have questions

Reply #28 –
I want to rectify something back in the thread. When I said that this:

Quote
As for my concern, do you think Opus 192 VBR is a little better than LAME V0 ?? I ask this because I saw many of your posts in which you catch some problem samples.

I actually meant shadowking, not halb27.

Thanks for halb27's explanation on bit reservoir and making the statement why software does not show that side of LAME.

When I am asking which codec is superior, it is purely on technical terms, as of: which is the one with few or virtually none killer samples, with high bitrates. It's is understood by the 2014 listening test, all 4 codecs used (excluding low anchor) are exceptionally good above 128 kbps.

Re: Looking at Opus for MP3 replacement and have questions

Reply #29 –
Quote from: halb27 on
@david-lisb:
Your wishes for a higher cutoff frequency can easily be fulfilled with Lame and lame3995o by using the - - lowpass option.
But if you prefer Ogg Vorbis: Go ahead.
FWIW, I'm not buying that any of the reasoning in david-lisb's post complies with TOS8.  I'd call bullshit, but he isn't the OP and it would be a distraction.  Besides, the OP knows better.
Is 24-bit/192kHz good enough for your lo-fi vinyl, or do you need 32/384?

Re: Looking at Opus for MP3 replacement and have questions

Reply #30 –
Quote from: halb27 on
Currently we don't know much about Opus'  quality at higher bitrates ...
:-X  :D
This is the most unbeleivable excuse/argument I've ever heard lately. It really is.

There were varios reports of very good listeners  of how Opus performs on high bitrates.  Just in case if someone don't remeber.
https://hydrogenaud.io/index.php/topic,108211.msg888651.html#msg888651

Otherwise feel free to show where Opus doesn't perform well at high bitrates. There was somebody talking about Opus's problematic perfomance on high bitrate few years ago. We're still waiting for samples from that person.

P.S. And Opus 96 kbps hits high 4.653 score in the last public test.  What?  "4.653 it's not enough" but LAME's score 4.237 is enough for you.
OK, 64 kbps wasn't enough high bitrate where Opus has hit score 4.0 . Then we tested  96 kbps where it hits 4.653.  "But we still don't know how it performs at 128k...". Are You telling that You're not sure that Opus won't hit 4.7-4.8 at 128 kbps which is on par with  ITU definition of "transparency" which was used 20 years ago in performance verificartion test of AAC 128k ?

Re: Looking at Opus for MP3 replacement and have questions

Reply #31 –
IgorC,
I really don't like to argue with you forever. So this is my last post on this aspect.

I said that we don't have a lot of information about Opus quality at high bitrates. High bitrate means significantly higher than 128 kbps to me. You point to 2 personal listening tests with higher bitrates.
Jplus' test contains 8 problem samples, and his result is 'If I were to rank the codecs for their performance at high quality settings, QT AAC ends up at a distinct first place and Opus at second place, while I'd have a hard time to decide whether to put LAME or aoTuV next. ... Opus was judged fully transparent at VBR with target bitrate 224kbps, which is consirably higher than I expected based on previous reports. At preset 192 I judged it untransparent so there's no grey area like in AAC or Vorbis. Opus VBR seems to be a lot less variable than the other codecs so in CBR mode I would trust Opus files of 230kbps and up.'
Steve Forte Rio tested 4 problem samples and the average score was 5 for Opus and 4.6 for QAAC.

So we have two personal high bitrate tests, and they have a different outcome (no miracle of course with different listeners for different samples).

This is not exactly what your post is suggesting,

I never said Opus isn't a great codec, and whoever wants to use it with higher bitrate: it certainly isn't unwise.
There's just no evidence that is to be prefered over AAC, especially as AAC is a no-brainer on the playback side. Exactly this is my point, not scepticism about Opus quality.

As for me personally (as you are again adressing this though this is a totally different story): Probably Opus quality at very moderate bitrate is good enough for me (never tested it, sure I care about tonal samples). But as AAC and mp3 quality is also good enough for me at bitrates I can easily tolerate (I'm not into -V5 like you are suggesting) I have no motivation to try Opus. I can play my mp3 library everywhere, share mp3 tracks with family and friends without problems, and I love having things easy.

Sure everybody does it the way he likes. You're a great admirer of Opus obviously. So use Opus.
lame3995o -Q1.7
opus --bitrate 140

Re: Looking at Opus for MP3 replacement and have questions

Reply #32 –
Quote from: halb27 on
You're a great admirer of Opus obviously.
And big admirer of LC-AAC, HE-AAC, xHE-AAC/USAC, Vorbis, LAME MP3, Helix, EVS, 3DAudio, MPEG2, MPEG4 ASP XviD, AVC/H.264, HEVC, Daala, Theora,  VP3/VP4/VP5/VP6/VP7/VP8/VP9/AV1.
And a tester of big-name cormmercial AAC and xHE-AAC encoders.
And?


Speaking of MP3 and AAC wide compability, yes, it was a panacea several years ago when users had to stick with what firmware support. Today most of people use mobile devices with Android, iOS, Windows. Heck, even foobar2000 now has Android and iOS version. People's preferements are changing:
HA's polls. I know they aren't 100% representative what happens globally but still.

Re: Looking at Opus for MP3 replacement and have questions

Reply #34 –
I think the reasonable takeaway here is people should not blindly assume that Opus  at 320 is going to be better than aac at 320 or mp3 at 320 for that matter.

Arguments from ignorance are just that: logical fallacies.
Is 24-bit/192kHz good enough for your lo-fi vinyl, or do you need 32/384?

Re: Looking at Opus for MP3 replacement and have questions

Reply #35 –
OK, a little bit of information theory here... One property of all (not totally stupid) quantizers is that their noise does down by 6 dB every time we increase the bitrate by one bit/coefficient. The difference between a good quantizer and a bad one is how many bits it takes to reach a fixed low-quality level. Same with different predictors, entropy coders, ... So while there may be a large difference between codecs at low bit-rates, like 48 or 64 kb/s, once you reach a high-enough quality point (around 96-128 kb/s), the rate at which quality improves as a function of bit-rate tends to be constant across codecs.

Looking at the results of the latest 96 kb/s listening test, I would say that the difference between Opus and Vorbis was probably equivalent to about 10 kb/s, with AAC being right in the middle. Information theory says that barring significant differences in encoder decisions, the quality of Opus, AAC and Vorbis is going to increase at more or less the same rate. So once you reach 192 kb/s, you could expect the differences to still be equivalent to about 10 kb/s, which is now a ridiculously small difference. At that point, I won't claim that Opus is still exactly 10 kb/s better than AAC, but I will claim that whatever difference between the two (either way) will be really small and given how insanely hard it is to do quality testing at that rate, any test you will attempt will tell you that Opus, Vorbis and AAC are all statistically tied at that rate. Also, at that point, any difference you might hear with the original would be due to a mistake made by the encoder and not something related to the format. Really, if you want transparent quality (or anything above 128 kb/s), you can use pretty much anything and you'll be fine.

Now, I did not mention MP3 above. In the 96 kb/s test, 128 kb/s MP3 was about on par with 96 kb/s Vorbis, so MP3 is 32 kb/s worse. This means you would expect MP3 to require about 234 kb/s to match 192 kb/s Opus. That's probably not too far off, except for the some issues MP3 tends to have with transients at any bit-rate. So MP3 is probably the only codec for which 192 kb/s may not be on par with Opus/Vorbis/AAC.

Re: Looking at Opus for MP3 replacement and have questions

Reply #36 –
That would only hold true if all the formats allocated bits in a perfect manner and simply had different starting positions. 

However I believe that almost the entirety of format evolution is smarter and smarter (attempting to reach that perfect) allocation of bits, therefore expecting a consistent offset is incorrect.
Creature of habit.

Re: Looking at Opus for MP3 replacement and have questions

Reply #37 –
Quote from: Soap on
That would only hold true if all the formats allocated bits in a perfect manner and simply had different starting positions. 

However I believe that almost the entirety of format evolution is smarter and smarter (attempting to reach that perfect) allocation of bits, therefore expecting a consistent offset is incorrect.

The thing is that wherever you allocate the bits to sound good at 96-128 kb/s (which may differ slightly across codecs), the smart thing to do when you want to increase the bitrate is to pretty much evenly distribute the increase. That's what all the codecs do? Or have you made some psychoacoustic discoveries we haven't heard of?

Re: Looking at Opus for MP3 replacement and have questions

Reply #38 –
Quote from: Leo 69 on
The main reason I haven't switched to Opus is that I heard somewhere that it reconstructs high frequencies so that they are basically synthetic..like in HE-AAC, for instance. Is this a correct assumption or not?

Nah, it does not. It has a special "band folding" method of encoding high frequencies at low bitrates, but it always operates at 48 Khz. It does not bluntly downsample it to 22 Khz and then fake half of the frequency spectra. The downside is that under 48 kbps it sounds more "cassetty" and muddy than HE-AAC, through in my opinion less artificial (HE-AAC sounds metallic to me at low bitrates).

Re: Looking at Opus for MP3 replacement and have questions

Reply #39 –
Quote from: Neuron on
The downside is that under 48 kbps it sounds more "cassetty" and muddy than HE-AAC, through in my opinion less artificial (HE-AAC sounds metallic to me at low bitrates).
You'll probably agree with me (and with TOS #8 ) that statements such as these would only make sense if backed up by ABX tests results.
Other than that they're not helpful at all, are they? Not even to justify the nechroposting.
Listen to the music, not the media.

Re: Looking at Opus for MP3 replacement and have questions

Reply #40 –
No, I do not. ABX testing would only prove that I can hear difference, not the nature of it, and I think it is fairly clear that at 48 kbps each codec has their own artifacts. No codec is transparent at 48 kbps yet, look at listening tests.

Re: Looking at Opus for MP3 replacement and have questions

Reply #41 –
ABC/HR would be the correct tool in this case.
 

Re: Looking at Opus for MP3 replacement and have questions

Reply #42 –
Quote from: Neuron on
No, I do not. ABX testing would only prove that I can hear difference, not the nature of it, and I think it is fairly clear that at 48 kbps each codec has their own artifacts. No codec is transparent at 48 kbps yet, look at listening tests.
No siree! Subjective, arcane statements such as "cassetty", "muddy" or "metallic", used by you in such a context, only belong into the typical audiophile forum.

Were you willing to follow the afore-mentioned science-driven method (which, as you know, and thank god, makes up one of the pillars of this community) a simple "Yes I can/No, I can't (hear differences)" should suffice. As simple as that.

Other than that, your claims stand as they are: discredited and of no use to anyone else.
Listen to the music, not the media.

Re: Looking at Opus for MP3 replacement and have questions

Reply #44 –
The move has also created somewhat of a shitstorm, with a bunch of people complaining loudly about it. Most of them seem to be stuck in a late-90s/early-2000s "more bitrate = always more better" mindset of 64kbps Xing-encoded MP3s downloaded from Napster, and don't seem to realize that as lossy formats get better, the required bitrate for transparency gets lower. We've had 20 years of improvements since then, which they also seem to ignore.

I find it very hard to ABX ~64kbps Opus against lossless and the default setting of ~96kbps is 100% transparent for me, no matter which samples I use. I can reliably ABX LAME -V5 MP3s (~128kbps), so for me Soundcloud's move should absolutely improve the sound quality.

I did hear some complaints that they used an old encoder version, though. My comparisons were done using opusenc v1.2.

Re: Looking at Opus for MP3 replacement and have questions

Reply #45 –
Quote from: KozmoNaut on
The move has also created somewhat of a shitstorm, with a bunch of people complaining loudly about it. Most of them seem to be stuck in a late-90s/early-2000s "more bitrate = always more better" mindset of 64kbps Xing-encoded MP3s downloaded from Napster, and don't seem to realize that as lossy formats get better, the required bitrate for transparency gets lower. We've had 20 years of improvements since then, which they also seem to ignore.

I find it very hard to ABX ~64kbps Opus against lossless and the default setting of ~96kbps is 100% transparent for me, no matter which samples I use. I can reliably ABX LAME -V5 MP3s (~128kbps), so for me Soundcloud's move should absolutely improve the sound quality.

I did hear some complaints that they used an old encoder version, though. My comparisons were done using opusenc v1.2.
The thing is that these improvements do not apply only to one specific format.
It's not like mp3 the last 20 years was a dead format that received no quality improvements. We cannot compare the 2000s mp3 encoders at 128kbps with Soundcloud's Lame-based 128kbps encoder that is up to date. For alot of people 128kbps can be transparent or very close to that, while a 64kbps opus file may not be far from transparent but I find it a stretch to say that it's as close to transparent as 128kbps lame is. I agree that they shouldn't have gone less than 80kbps. The move they did was just because they wanted to save a set target amount bandwidth with as little as possible loss in quality. Opus was definitely the answer. That doesn't mean that they didn't make a significant compromise in quality though, it just means it was in their least important thing out of a list.

Re: Looking at Opus for MP3 replacement and have questions

Reply #46 –
Quote from: Klimis on

It's not like mp3 the last 20 years was a dead format that received no quality improvements. We cannot compare the 2000s mp3 encoders at 128kbps with Soundcloud's Lame-based 128kbps encoder that is up to date.

Lame 3.93 came out in 2002, and while there was some improvements over the next 4 or 5 years, 2000s mp3 encoders were not that different from the ones we have today.

Re: Looking at Opus for MP3 replacement and have questions

Reply #47 –
Quote from: saratoga on
Quote from: Klimis on

It's not like mp3 the last 20 years was a dead format that received no quality improvements. We cannot compare the 2000s mp3 encoders at 128kbps with Soundcloud's Lame-based 128kbps encoder that is up to date.

Lame 3.93 came out in 2002, and while there was some improvements over the next 4 or 5 years, 2000s mp3 encoders were not that different from the ones we have today.
Still they is no comparison with any generic mp3 encoders of that time of most applications that cared little about quality at lower bitrates (like most music library programs that had a rip to mp3 function). I clearly remember struggling to encode something below 192kbps without the sound starting to get distracting on my tiny storage space phone. Most generic encoders of that time had a nose dive in sound quality below 192kbps. Lame offered unparalleled audio quality and still does for what mp3 is capable of and it still managed to pull the boundaries atleast for a slight bit the last decade or so. I wish I knew about it then, but then I wish I knew that my phone would play AAC-LC and I knew the technical advantages, but even more I wish there was opus around, it would have saved me hours of exchanging tracks on my phone's storage depending on my mood for music for the day.

Re: Looking at Opus for MP3 replacement and have questions

Reply #48 –
Quote from: Klimis on
For alot of people 128kbps can be transparent or very close to that, while a 64kbps opus file may not be far from transparent but I find it a stretch to say that it's as close to transparent as 128kbps lame is.

It is equally good for me, ie. no noticeable artifacts or degradation, unless I listen carefully in direct comparison with a lossless version.

That's what I can say with confidence. Other people may be more sensitive to any artifacts and such.

Re: Looking at Opus for MP3 replacement and have questions

Reply #49 –
The user most vocal about the decrease in sound quality in soundcloud even wrote an article about it Opus Tests. I'm not sure though if his findings are accurate.
 
SimplePortal 1.0.0 RC1 © 2008-2018