Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: best converter for xhe aac encoder (Read 13077 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Re: best converter for xhe aac encoder

Reply #25
I am trying to figure out which is better as well, opus or xhe-aac.  I read from this forum saying that xhe aac is better than opus in low bitrate
Opus has better compatibility with smartphone players and it's almost 3x times faster in terms of decoding so you get better battery life. I guess that's what matters at this bitrate because both codecs are not consistently better than the other in terms of sound quality.
Really?  i guess i would stick with opus then, thanks!

Re: best converter for xhe aac encoder

Reply #26
Opus has better compatibility with smartphone players and it's almost 3x times faster in terms of decoding so you get better battery life. I guess that's what matters at this bitrate because both codecs are not consistently better than the other in terms of sound quality.

It's been 24 hours now and he's probably not answering you to keep himself decent. Opus doesn't work on iPhone, that's no small thing when you consider what music listeners own when they can't add a microSD card. To be precise it is not true that Opus does not work but rather that the format war has now moved to their containers and should rewrite everything in CAF, notoriously iPhone does not read Ogg and Webm. It is not even true that he will save battery because to listen to his music in headphones at the time of the disappearance of the jack will have the phone in perennial re-encoding for Bluetooth headphones.

Compared to xHE-AAC the sound of Opus is more closed, less brilliant, even with the human voice alone you can perceive the disappearance of the treble at low bitrates.

Alternatively, you may believe that those who prefer to pay for a better encoder are fools.

Re: best converter for xhe aac encoder

Reply #27
Opus doesn't work on iPhone,
It does if you use a player like VLC or foobar2000 at least.

When you talk about CAF, you talk about Opus being used in browsers.

Compared to xHE-AAC the sound of Opus is more closed, less brilliant, even with the human voice alone you can perceive the disappearance of the treble at low bitrates.
It's exactly the opposite to what you say if we talk about my experience. xHE-AAC is less trebly on lower bitrates, and that's its strength sometimes - as I said, I can't give a perfect win to either codec, Opus is better at saving HF but sometimes creates more artifacts just because of that, so the result depends on source material.

Alternatively, you may believe that those who prefer to pay for a better encoder are fools.
It's not about money, xHE-AAC is aimed at different markets.
Opus VBR 256 + SoX

Re: best converter for xhe aac encoder

Reply #28
No, I am not speaking at the browser level (it is a problem of current operating systems that decode Opus in Webm only at the browser level). I'm talking about iOS, if the decoder (and the container too) is implemented by the operating system it will be available for every app.

My contents are normalized to -23dBFS (EBU R128), on the treble we talk about perception, the Opus encoder exploits our poor resolving capacity so it can happen with loudness war contents to perceive the opposite.

I propose a test to prove the problem: get a file containing white noise, at each frequency you will have exactly the same noise level in absolute values.

Open the terminal and ask sox to generate the spectrogram for you:
Code: [Select]
sox file.wav -n spectrogram -o wave.png

You will get something like the following:
X

Now compress the file with Opus:
Code: [Select]
opusenc file.wav file.opus

and decompress it again as Wave:
Code: [Select]
opusdec file.opus opus.wav

Now ask the Sox for a new spectrogram and admire the nonexistent bit depth reserved for the treble:
Code: [Select]
sox opus.wav -n spectrogram -o opus.png

The result will look like the following:
X

I hate HE-AAC, therefore until now I serve the contents I produce to Windows as Opus in Webm and therefore I am absolutely sure of what I write on the bandwidth, at 36kbps a mono file containing only voice is reproduced by Opus in an excellent way with cutting higher frequencies. Should I make the comparison in stereo? I guess not, you will have to do this with music, but I already know that xHE-AAC offers the best on vocals.

If Opus were superior to xHE-AAC for in newest TV standard and DRM radio we found Opus instead MPEG-D USAC, broadcast contents are mainly vocal and therefore the advantage of one of the two emerges, although Opus is the best opensource lossy encoder I have ever tried.

Re: best converter for xhe aac encoder

Reply #29
Open the terminal and ask sox to generate the spectrogram for you:
I do that in Audacity and I know what you're talking about. My spectrograms do not look like yours for Opus.



Although there certainly IS something happening at 15.6 khz, it's not as tragical as on your spectro. It's not even as tragical as MP3 16 khz half-cutoff due to sfb21. Btw bitrate is 64 kbps on this picture.

But I was talking strongly about listening experience - Opus is better at preserving energy at high frequencies than xHE-AAC on the material I tried. I actually do not even know why, because xHE-AAC spectro doesn't show anything shabby about it, but I hear it during ABX at 96 kbps (exhale 3). It's even more pronounced at 64 kbps, but Exhale cuts frequencies above 14.5 khz there so this may be the reason as well (Opus is crazy enough to do 20 khz even at this bitrate).

If Opus were superior to xHE-AAC for in newest TV standard and DRM radio we found Opus instead MPEG-D USAC
Let me repeat that once again - xHE-AAC is aimed at different markets. The decision to use xHE-AAC in new TV and Radio standards has nothing to do with the sound quality of Opus, it's a business and political decision where Opus didn't have a chance exactly because it's royalty free and maintained by organizations who are not particulary strong at lawsuits.

You could guess why Opus haven't and will never be used in Bluetooth. It's a perfect codec for this usecase, because it's very low latency and it provides high quality at low bitrates comparing to any other BT codec on the market. Why isn't it used there? For the very same reason - because it's open source, royalty free and it's not by Fraunhofer, Dolby, Qualcomm or some other huge name. Simple as that.

Things are better in video codec market because AOMedia and companies there are aggressive enough to promote AV1 just everywhere. Maybe some day there will be an Opus successor that will meet the same destiny.
Opus VBR 256 + SoX

Re: best converter for xhe aac encoder

Reply #30
But I was talking strongly about listening experience - Opus is better at preserving energy at high frequencies than xHE-AAC on the material I tried. I actually do not even know why, because xHE-AAC spectro doesn't show anything shabby about it, but I hear it during ABX at 96 kbps (exhale 3). It's even more pronounced at 64 kbps, but Exhale cuts frequencies above 14.5 khz there so this may be the reason as well (Opus is crazy enough to do 20 khz even at this bitrate).
I don't plan to enter in this spectrograph debate, but I'd like to inform you that Exhale has a SBR mode. If you try preset c instead of preset 3, higher frequencies are also coded.

I also want to add than neither OPUS nor any transform codec preserves energy as you say: they are transform codec, and energy is rather recreated bu tools and algorithms than preserved. This recreation is always an approximation. It becomes really obvious with modern encoders at low bitrate. You can both hear higher frequency and how distorted they are. xHE-AAC is crazy enough to go to 15 KHz at 12 kbps (Fraunhofer's implementation).

I finally get into this debate… :-*


Re: best converter for xhe aac encoder

Reply #32
I can think of other things like re encoding headroom or if you dont trust the
encoder yet at lower settings or its VBR..
Thing is, you cant even trust em at high bitrates. There is no quality guarantee with lossy.

You can reach a level considered transparent or not annoying on standard even electro music.

Re: best converter for xhe aac encoder

Reply #33
I don't plan to enter in this spectrograph debate, but I'd like to inform you that Exhale has a SBR mode.
That's a great feature indeed, but I personally don't really like what SBR does with the sound.

I also want to add than neither OPUS nor any transform codec preserves energy as you say: they are transform codec, and energy is rather recreated bu tools and algorithms than preserved.
What I was trying to say is that Opus doesn't alter the level of overall HF presence to my personal hearing even at 64 kbps - high frequencies are at the same level as they are in the source, and it does that without SBR. With xHE-AAC there's another story though - I hear it starting to tame HF at 96 kbps and below. I wouldn't say it's necessarily a bad thing, because the codec sounds somewhat cleaner and less distorted at this bitrate during some samples, but I was answering to the claim that xHE-AAC sounds "more spacious, sparkly, defined, etc", which is definitely not true in this case.

 Maybe it's exclusively an Exhale thing - applying a slow roll-off lowpass filter for example. I have only used Exhale)
Opus VBR 256 + SoX

Re: best converter for xhe aac encoder

Reply #34
What I was trying to say is that Opus doesn't alter the level of overall HF presence to my personal hearing even at 64 kbps
All lossy encoders have problems for this reason they only make sense to save space, not to store content, the problem with Opus is that if the signal is missing, whether it is a hole (or drop) or a frequency limited source it will fills with noise. Under normal conditions this noise is masked.

Try for a moment to think what would happen to an uncompressed file where the high frequencies are encoded with 5 or less bits, what would happen? It would be full of noise whether your ears hear it or not. I'm not starting from the spectrogram.

Why can't you find many podcasts in Opus? They are almost all in MP3 which has roughly the frequency response of a dynamic microphone (see https://www.shure.com/fr-FR/produits/microphones/sm58?variant=SM58-LCE
Réponse en fréquence: 50 à 15.000 Hz). I could generate white noise with Sox, I searched for it in MP3 to better show the problem. The solution is simple, just input a 32kHz sampled signal and there won't be any noise added above 16kHz.

I purposely used the default settings. The thesis that Opus was not chosen for broadcast programs for political reasons is nonsense, when I have time I will explain why. Neither Opus nor Exhale have a market, my aim was just to show a known problem (since 2017 in this forum).
X


Re: best converter for xhe aac encoder

Reply #36
Also if you're interested - I cannot replicate what you try to claim about Opus on white and pink noise signals as well (btw your spectros look more like a pink noise, not white). No dependance on signal volume as well.
Opus VBR 256 + SoX

Re: best converter for xhe aac encoder

Reply #37
Sorry, I can't upload the wave file for the size.

So I start all over again generating the band limited white noise with Sox:
Code: [Select]
sox -n -r 32k whitenoise.wav synth 30 whitenoise

At this point we have all the high frequencies that you can record with the most used microphone in the world for live performances.

Code: [Select]
ffmpeg -i whitenoise.wav -ar 48k input.wav

At the mixer output we will have a 48kHz track ready to be broadcast. If we compress a file without signals with Opus we will obtain a completely black spectrogram. If, on the other hand, we provide a limited frequency signal, such as the one the Shure SM58 microphone would provide, Opus will shift a bad copy of what it finds in the previous bands to high frequency and fill the entire audible high frequency band with noise.

 

Re: best converter for xhe aac encoder

Reply #38
Code: [Select]
ffmpeg -i whitenoise.wav -ar 48k input.wav
This is the most exploited shit I have ever seen on this forum. You're asking Opus to save a perfect copy of a badly resampled white noise at a low bitrate? Really?

And no, that is not representation of human voice through the mic by any means.
Opus VBR 256 + SoX