Author Topic: opus, aac or something else?  (Read 22646 times)

Ansem

  • Jr. Member
  • **
  • Posts: 99
Hi everyone!
I usually rip my CDs in flac since it's loseless, but i wandered what files format will come out now that mp3 should be dead (and I hope so since it's terribile).
I read very well of opus that is open source and comes with a great quality (up to 500kbps). Do you think that it will became the next standard? Or aac is already too famous?

There are any good encoder for them?

theta_wave

  • Sr. Member
  • ****
  • Posts: 680
Why should MP3s be dead?  The format is widely-supported, mp3 decoding is no longer patent encumbered and their tag spec (id3) is completely rational compared to the horrid m4a spec (the container of choice for aac).  I highly doubt LAME encoded MP3s using the V0 preset would be distinguishable from AAC 256kbps under ABX.  Basically, Fraunhofer IIS is pushing this "mp3 is dead" carnard because it can no longer make money off of its recently expired MP3 patents, while its AAC patents are still in full effect.

As for OPUS, upsampling FLACs from CDs (sampled at 44.1kHz) to 48kHz doesn't seem ideal to me but OPUS supporters would claim that OPUS resamplers are good.  I haven't messed with OPUS yet nor do I feel compelled to due to the format's lack of widespread default support among different OS's and devices.


GoodGuy

  • Newbie
  • *
  • Posts: 5
  • Full time webmaster without hobbies ;)
I  know this thread is old but I found out about Opus a couple of weeks ago and it's amazing!

I've been converting 320/kbps MP3s (I know it's not ideal to convert from lossy formats) and also from lossless formats like FLAC into 96-128/kbps Opus files just to save space on my hard drives.

I know my computer speakers (2.1) isn't the best, people with really good equipment will probably hear a difference but going from MP3s that's ~12 MB into songs that's ~3.5MB. It's a huge difference if you have thousands of them and there's still not a major difference in quality, at least for my ears.

Also, it only takes about 10 seconds to convert a song (with LameXP).

I've also started to use Opus for audiobooks (you can really push the bitrate way down here) and also when ripping movies I now only use x265 and Opus.


Chrome, Firefox and Edge does support Opus.
Youtube started using it too in 2014.
Android 5.0+ supports it.
VLC.
.........



BeeBeeKing

  • Full Member
  • ***
  • Posts: 236
Amazing, but i'll be staying with V0 preset mp3!  8)

frankz

  • Sr. Member
  • ****
  • Posts: 3835
I notice a lot of forums where people register and then only make posts pushing this Opus.  It's very odd.  Either it's great and has a devoted cult-like following, or someone is making money off of it.

Either way, mp3 is universal and established and not going anywhere any time soon.  It's going to be FLAC for lossless and mp3 for lossy/portable for a long time coming, probably our lifetimes.

GoodGuy

  • Newbie
  • *
  • Posts: 5
  • Full time webmaster without hobbies ;)
I notice a lot of forums where people register and then only make posts pushing this Opus.  It's very odd.  Either it's great and has a devoted cult-like following, or someone is making money off of it.

Either way, mp3 is universal and established and not going anywhere any time soon.  It's going to be FLAC for lossless and mp3 for lossy/portable for a long time coming, probably our lifetimes.

It's a conspiracy! ;)

No, the thing is that I've used MusicBee for years and I used to have a account here, at least according to the plugin Lastpass, but couldn't login (too long ago I did that I assume so my acc. got deleted?) so I did register again just to answer this post I found through Google.

I don't develop Opus or have anything to do with it other than recently discovered that it's really good at compressing the audio.

And you can say that I promote it in a way because I really want more people to use it.
That way all these release groups (of legal material of course) could use higher bitrate for the video since Opus don't need much for high quality audio.
And music albums and such would also have a smaller file size of course but still the same good quality.

Just my 2 cents. Have a nice day guys!  8)

sveakul

  • Sr. Member
  • ****
  • Posts: 2438
I think it boils down to personal preference, like so many things.  For me, the goal is a codec that sounds transparent to the source with multiple music types to my gettin' old ears, and given that, produces the smallest file size.  That's been Opus at 160k for me.  I have lots of mp3 files and wouldn't think of transcoding them, but do convert FLAC files and rip my CDs to Opus.  I agree about mp3 remaining much more device-compatible, and that V0 VBR stands up to any listening test.

frankz

  • Sr. Member
  • ****
  • Posts: 3835
I was curious about opus right up until the point where I read that, at any bitrate that would provide any significant space savings over V0 MP3, it synthesizes high-end frequency data. Same thing with AAC.  No thanks.  I'll take a few wasted bits over invented audio data that's not accurate to the source by design.

I've tried all these codecs as they come along, but I've been using MP3 since 1999 - the Xing / Napster days - and mp3 has improved and evolved and seen universal adoption while all the others have faded into obscurity.  And this despite the fact that, for the longest time, real mp3 was locked down by patent, and all free versions were hacked out emulators.

hiccup

  • Sr. Member
  • ****
  • Posts: 7790
Only because I was a bit curious about the technicalities and possible advantages on this matter;

If I am correct, even the developers of Opus advertise that any benefits are pretty much only there when encoding to 128kbps or lower.
So that would be in situations where you are seriously challenged for space or bandwidth.

Out of curiosity about resulting file sizes (for higher quality encoding) I just encoded a 50 min. flac file to both mp3 and opus (approx. 160kbs)
The resulting file sizes were respectively 63.710 and 63.985 KB.
So for any 'serious' music listening I can't see any advantages to using opus over mp3?

The only aspect that might be an advantage, is that opus is using (if I am correct) Vorbis comment for it's metadata?
Would that be of any benefit in real-world (MusicBee) usage?

frankz

  • Sr. Member
  • ****
  • Posts: 3835
Of course if you encode at the same bitrate you're going to get the same approximate file size.  Bits per second translate into bytes and megabytes per minute no matter what codec it is. That's math.

160 kilobits per second = .02 megabytes per second = ~60 megabytes for 50 minutes.  Then you have some tagging and file overhead on top.

The supposed benefit is that opus will allegedly sound better at 160 kpbs than mp3. Which means to get the sound quality of a 256 kpbs MP3, you can theoretically use a lower bitrate for opus, resulting in smaller sound file for the same quality.  Maybe a 256 kbps mp3 would sound equal to a 192 kbps opus file to you (for example), and you could save those 0.5 megabytes per minute in space.  That's the theory, anyway.

My problem is I'm not a fan of how they get there (inventing data), and space is cheap enough that it's worth it to use what's compatible with everything rather than trying to reinvent the wheel for half a megabyte per minute.
Last Edit: November 18, 2017, 07:43:46 PM by frankz

hiccup

  • Sr. Member
  • ****
  • Posts: 7790
Of course if you encode at the same bitrate you're going to get the same approximate file size.  Bits per second translate into bytes and megabytes per minute no matter what codec it is. That's math.

I don't think that's actually true?
E.g. 160kbps only indicates the data transfer rate, and doesn't directly relate to the size of the source file.
Depending on the efficiency and complexity of the de- encoder, and the contents of the audio file, the file size and the data rate will not be (easily) mathematically related.

Are there reliable abx tests that prove that opus can sound better than mp3 at bitrates above 160kbps?

Even the opus developers themselves seem to be 'making their case' predominantly for <128 kbps:


frankz

  • Sr. Member
  • ****
  • Posts: 3835

I don't think that's actually true?
E.g. 160kbps only indicates the data transfer rate, and doesn't directly relate to the size of the source file.
Depending on the efficiency and complexity of the de- encoder, and the contents of the audio file, the file size and the data rate will not be (easily) mathematically related.
So your two files just happened to be less than 300 b different across 50 minutes?

No.  A second of audio is a second of audio both stored and transferred.  CDs have a bitrate of 1411 kbps (0.176375 Megabytes per second), and a CD-ROM with 847 MB capacity can hold 80 minutes of music (0.176375 * 60 * 80 = 846.6).  The 1411 kbps is also math - 44100 samples * 16 bits * 2 channels = 1,411,200 bits)

Are there reliable abx tests that prove that opus can sound better than mp3 at bitrates above 160kbps?

Even the opus developers themselves seem to be 'making their case' predominantly for <128 kbps:


I believe they do, but you'd have to ask one of its boosters for details.  From what I read, they not only claim that it sounds better at low bit rates but also at each point as the scale goes up (and beyond mp3's limit).  It somehow achieves this while degrading everything by converting 44100 to 48000 for no earthly reason.
Last Edit: November 18, 2017, 10:51:12 PM by frankz

hiccup

  • Sr. Member
  • ****
  • Posts: 7790

I don't think that's actually true?
E.g. 160kbps only indicates the data transfer rate, and doesn't directly relate to the size of the source file.
Depending on the efficiency and complexity of the de- encoder, and the contents of the audio file, the file size and the data rate will not be (easily) mathematically related.

So your two files just happened to be less than 300 b different across 50 minutes?

No, I don't think they just happen to be near identical in size.
I think it is because of the efficiency of the encoding algorithms of opus and mp3 being similar.

To show my point: some mp3 encoders have an 'hq' setting.
That will not influence the set bit-rate, but it will (theoretically) give a higher sound quality.
And a slightly larger file size...

Flac is an even clearer example.
You can set it to use different compression levels.
All resulting in the same bit-rate output, but with different file sizes.
(but also pretty much negligible in practice)

I am sure more space efficient lossy audio algorithms could be developed (or already exist).
But the minor gains in storage space will probably not weigh up against the more computing power needed. (battery life of portable players)

The upsampling to 48khz that opus does, is motivated by the fact that cheaper dacs have some audio degradation when playing at 44.1khz.
(there is some factual and justifiable math behind that)
And they believe that the degradation from upsampling to 48khz is preferable over the sound deterioration that occurs with these cheaper dacs playing at 44.1khz.

So again, opus seems to be mainly targeted at low to medium audio quality demands.
And for streaming purposes. It is said to be designed to be more robust compared to other lossy codecs.
(better latency behavior)

So it seems to be safe to conclude that there are no good reasons to use opus at higher bitrates on 'good' and modern equipment.

frankz

  • Sr. Member
  • ****
  • Posts: 3835
Flac is an even clearer example.
You can set it to use different compression levels.
All resulting in the same bit-rate output, but with different file sizes.
(but also pretty much negligible in practice)
Well, FLAC is variable bit rate, which is not what we're talking about here, but even on its face this claim is decidedly not true.   The definition of compression is changing the bit rate.  

Here is Asia's Heat of the Moment compressed with two different FLAC schemes for example.

Level 4


and Level 8


Decreasing the bit rate while maintaining the information contained in the original bits is how lossless compression works.

To show my point: some mp3 encoders have an 'hq' setting.
That will not influence the set bit-rate, but it will (theoretically) give a higher sound quality.
And a slightly larger file size...
Which mp3 encoder are you talking about here?

If you can encode the same audio file with the same encoder and the same bit rate in constant bit rate mode and get different sized files, I'd love to see evidence.
Last Edit: November 19, 2017, 02:41:53 PM by frankz