HydrogenAudio

Lossy Audio Compression => AAC => Topic started by: KeithPhantom on 2016-01-20 00:44:51

Title: xHE-AAC vs HE-AACvE
Post by: KeithPhantom on 2016-01-20 00:44:51
I recently saw the new codec, xHE-AAC an I've some questions:

1) Do yo think that this codec can be better than HE-AACv2?

2)Can this codec offer better quality at lower bitrates than HE-AACv2?

3) How many encoders and decoders are there available now?

4)Will the smartphones play this format (I use AAC-LC VBR 256 kbps with FAAC encoder and compresed from the lossless and played with Blackberry 10 OS)? It's HE AACv2
Title: Re: xHE-AAC vs HE-AACvE
Post by: pdq on 2016-01-20 01:14:07
https://hydrogenaud.io/index.php/topic,109486.0.html
Title: Re: xHE-AAC vs HE-AACvE
Post by: LithosZA on 2016-01-20 07:05:56
Quote
1) Do yo think that this codec can be better than HE-AACv2?

2)Can this codec offer better quality at lower bitrates than HE-AACv2?
From the listening test performed it seems like xHE-AAC only has an advantage at bitrates smaller than 32Kbps.

Quote
3) How many encoders and decoders are there available now?
Nothing for the normal consumer at the moment.

Quote
4)Will the smartphones play this format (I use AAC-LC VBR 256 kbps with FAAC encoder and compresed from the lossless and played with Blackberry 10 OS)? It's HE AACv2
Not sure, but you can get a quality increase by simply switching to different LC-AAC encoders such as the Apple or Fraunhofer AAC

Title: Re: xHE-AAC vs HE-AACvE
Post by: KeithPhantom on 2016-01-20 18:14:25
Quote
1) Do yo think that this codec can be better than HE-AACv2?

2)Can this codec offer better quality at lower bitrates than HE-AACv2?
From the listening test performed it seems like xHE-AAC only has an advantage at bitrates smaller than 32Kbps.

Quote
3) How many encoders and decoders are there available now?
Nothing for the normal consumer at the moment.

Quote
4)Will the smartphones play this format (I use AAC-LC VBR 256 kbps with FAAC encoder and compresed from the lossless and played with Blackberry 10 OS)? It's HE AACv2
Not sure, but you can get a quality increase by simply switching to different LC-AAC encoders such as the Apple or Fraunhofer AAC



I don't use the Nero AAC because I really don't like the .m4a extension file.
Title: Re: xHE-AAC vs HE-AACvE
Post by: xnor on 2016-01-20 18:22:59
If you want better quality then switch to a better encoder, like LithosZA suggested.

The file extension is completely irrelevant. You can simply rename m4a to mp4.
Title: Re: xHE-AAC vs HE-AACvE
Post by: saratoga on 2016-01-20 18:23:46
I don't use the Nero AAC because I really don't like the .m4a extension file.

You mean you don't like AAC files?  Or are you literally saying you want to rename the files to use a different extension?
Title: Re: xHE-AAC vs HE-AACvE
Post by: xnor on 2016-01-20 18:34:02
You mean you don't like AAC files?  Or are you literally saying you want to rename the files to use a different extension?
Only the latter makes sense, since in #1 he wrote "I use AAC-LC VBR 256 kbps with FAAC encoder ".
Title: Re: xHE-AAC vs HE-AACvE
Post by: KeithPhantom on 2016-01-22 12:25:57
If you want better quality then switch to a better encoder, like LithosZA suggested.

The file extension is completely irrelevant. You can simply rename m4a to mp4.
I want to say that I like my AAC files with the extension .aac, no .m4a and no .mp4. I don't know if I can rename the extensions
Title: Re: xHE-AAC vs HE-AACvE
Post by: xnor on 2016-01-22 16:01:22
I want to say that I like my AAC files with the extension .aac, no .m4a and no .mp4. I don't know if I can rename the extensions
Of course you can.
Title: Re: xHE-AAC vs HE-AACvE
Post by: saratoga on 2016-01-22 16:17:22
Yeah, you can rename the file to whatever you want.  Some devices or programs may not like it though if you use a non-standard file extension though. 
Title: Re: xHE-AAC vs HE-AACvE
Post by: Zarggg on 2016-01-22 16:24:12
AAC files with the extension .aac are commonly containerless files that contain strictly AAC data. .mp4 and .m4a commonly refer to MPEG-4 Part 14 containers (https://en.wikipedia.org/wiki/MPEG-4_Part_14) which may hold AAC data or any of several audio/video streams (https://en.wikipedia.org/wiki/MPEG-4_Part_14#Data_streams).
Title: Re: xHE-AAC vs HE-AACvE
Post by: imacguru on 2017-10-23 18:33:40
I recently saw the new codec, xHE-AAC an I've some questions:

1) Do yo think that this codec can be better than HE-AACv2?

I think it's way better! Stereo separation is flawless compared to the half-way to stereo sound of HE-AACv2. The high-end is also much better thanks to enhanced SBR.


2)Can this codec offer better quality at lower bitrates than HE-AACv2?

This codec beats HE-AACv2 at any bitrate and especially at lower bit-rates. As mentioned, the parametric stereo is flawless and the highs are cleaner and clearer than HE-AACv2.


3) How many encoders and decoders are there available now?

Not many. I'm using the excellent StreamS Hi-Fi Encoders for my Internet Radio xHE-AAC streams and the StreamS Hi-Fi Radio app (only available for iPhone, iPad, iPod Touch and Apple TV) for decoding. The streams are 40kbps (average bit rate) HLS and to my ears sound awesome! I've found 40kbps to be the sweet spot for xHE-AAC where the audio is crystal clear. There may be other xHE-AAC encoders available but as far as I know StreamS is the best available. I'm running it on an HP PC in Windows 10 along with the Orban Optimod PCn1600 audio processing system and Virtual Audio Cables for internal audio routing.


4)Will the smartphones play this format (I use AAC-LC VBR 256 kbps with FAAC encoder and compresed from the lossless and played with Blackberry 10 OS)? It's HE AACv2

For now, the only app I know of that plays both HLS and Icecast xHE-AAC streams is StreamS Hi-Fi Radio and it's only available in iOS and tvOS. There are at this moment in time 12 Internet Radio stations in the app directory streaming xHE-AAC.

4 of the 12 xHE-AAC stations are mine. They are LG73.ca, MaxRadio.ca, NewWestRock.ca and UptownRadio.ca. I have other streams available for each station including HE-AACv1 and AAC since xHE-AAC isn't widely available (yet).

If you get a chance to hear them in StreamS Hi-Fi Radio, let me know what you think. I hope posting a reply 21 months later isn't a problem on this forum! I've become a big fan of xHE-AAC!


Cheers,
Phil
Title: Re: xHE-AAC vs HE-AACvE
Post by: kode54 on 2017-10-24 00:12:35
What even decodes xHE-AAC?
Title: Re: xHE-AAC vs HE-AACvE
Post by: IgorC on 2017-10-24 17:05:11
The high-end is also much better thanks to enhanced SBR.
eSBR doesn't bring noticeble improvements over standard SBR. If not we would see considerably better results at 64 kbps.
The selling point of xHE-AAC is improved AMR-WB+ coder (which actually is better for mono music and mono/stereo speech)  with improved parametric/unified stereo tools.  Just two main improvements: improved AMR-WB+ and unified stereo tools. Nothing more.

eSBR, new entropy coder and all others misc stuff just bring rose colors to main picture.

https://mpeg.chiariglione.org/sites/default/files/files/standards/parts/docs/w12232-v2-w12232.zip
MUSHRA scores at 64 kbps:
xHE-AAC - 81.9
HE-AAC  - 79.6

Let's put it in terms of Mean opinion score (MOS at 64 kbps):
xHE-AAC - 4.09
HE-AAC  - 3.98

Only +0.11 increment. And You should divide it between eSBR, improved unified stereo coding and improved entropy coder.

eSBR doesn't bring quality gain at 48-64 kbps. Its main task was to extend bandwidth at very low bitrates.
SBR extend BW by factor of 2 (2x) (22 kHz -> 44 kHz).  While eSBR has new factors of extension 2.67x and 4x.
Title: Re: xHE-AAC vs HE-AACvE
Post by: imacguru on 2018-03-16 02:00:06
eSBR doesn't bring noticeble improvements over standard SBR. If not we would see considerably better results at 64 kbps.
The selling point of xHE-AAC is improved AMR-WB+ coder (which actually is better for mono music and mono/stereo speech)  with improved parametric/unified stereo tools.  Just two main improvements: improved AMR-WB+ and unified stereo tools. Nothing more.

eSBR, new entropy coder and all others misc stuff just bring rose colors to main picture.

https://mpeg.chiariglione.org/sites/default/files/files/standards/parts/docs/w12232-v2-w12232.zip
MUSHRA scores at 64 kbps:
xHE-AAC - 81.9
HE-AAC  - 79.6

Let's put it in terms of Mean opinion score (MOS at 64 kbps):
xHE-AAC - 4.09
HE-AAC  - 3.98

Only +0.11 increment. And You should divide it between eSBR, improved unified stereo coding and improved entropy coder.

eSBR doesn't bring quality gain at 48-64 kbps. Its main task was to extend bandwidth at very low bitrates.
SBR extend BW by factor of 2 (2x) (22 kHz -> 44 kHz).  While eSBR has new factors of extension 2.67x and 4x.

Hi Igor! Apologies for the extremely delayed reply. I need to visit here more often.

You may recall in my post I mentioned that my 4 xHE-AAC streams are all 40kbps average bit-rate. My stations also have 64kbps streams which are all in HE-AACv1. The main point of this thread as I saw it was comparing xHE-AAC to HE-AACv2.

eSBR brings quality improvements (to my ears) at 40kbps, especially after the encoders I use were updated with Fraunhofer's latest generation xHE-AAC encoder which improved clarity in the high-end over their previous xHE-AAC encoder.

At 64kbps for compatibility and many other reasons I believe at this time HE-AACv1 is the best choice for encoding. It has excellent sound quality and stereo imaging. HE-AACv2 on the other hand due to its less than optimal parametric stereo is to my ears noticeably inferior to 64k HE-AACv1. xHE-AAC thanks to amazingly accurate parametric stereo sounds great at 64kbps. My own testing shows 40kbps is the sweet spot for great audio and efficiency!

The testing report you refer to is from September 2011 so it's outdated as Fraunhofer have made excellent progress on the xHE-AAC encoding/decoding since then.

If you have access to an iOS or tvOS device, I invite you to grab a copy of StreamS Hi-Fi Radio and check out my 40kbps xHE-AAC streams. Unless I'm in a state of self-delusion I think you'll be blown away by the audio quality! I've demo'd the streams to many people now and the response has been consistent with my own ears.


Cheers,
Phil
Title: Re: xHE-AAC vs HE-AACvE
Post by: Moni on 2018-03-16 13:36:42
imacguru, have you compared against the most recent Opus beta or development master? Lots of work has been done to improve Opus further at very low bitrates, including with parametric stereo. Opus has the added benefit of not being proprietary.
xHE-AAC had my interest when I first heard about it but the very limited availability/adoption combined with tremendous gains with Opus have me wondering if there is really much of a point to it.
I will check out your stream when I get a chance but I think it would be very instructive if you were to compare next to Opus as that is the real competition now, not HE-AAC.
https://hydrogenaud.io/index.php/topic,115156.0.html

Relevant: https://hydrogenaud.io/index.php/topic,115156.msg949927.html#msg949927
Title: Re: xHE-AAC vs HE-AACvE
Post by: imacguru on 2018-03-16 14:39:10
imacguru, have you compared against the most recent Opus beta or development master? Lots of work has been done to improve Opus further at very low bitrates, including with parametric stereo. Opus has the added benefit of not being proprietary.
xHE-AAC had my interest when I first heard about it but the very limited availability/adoption combined with tremendous gains with Opus have me wondering if there is really much of a point to it.
I will check out your stream when I get a chance but I think it would be very instructive if you were to compare next to Opus as that is the real competition now, not HE-AAC.
https://hydrogenaud.io/index.php/topic,115156.0.html

Relevant: https://hydrogenaud.io/index.php/topic,115156.msg949927.html#msg949927

Hi Moni!

Thanks for the tip to check out the Opus. The samples I listened to were 48kbps. I found consistent issues with the highs on the Opus files: several songs had excessive sibilance distortion. I noted a "metallic" effect on the highs which was not particularly pleasant to listen to. Stereo imaging was fine but that's not hard to do at 48kbps.

To my ears this is not even close to the quality and clarity of 40kbps xHE-AAC. It just doesn't sound hi-fi to me. For your reference I used my Bose QC 35 headphones for the listening tests.

Late last year Via Licensing changed the terms for xHE-AAC making it a free add-on for existing AAC licensees. Just a matter of time now until xHE-AAC becomes ubiquitous.


Regards,
Phil
Title: Re: xHE-AAC vs HE-AACvE
Post by: Moni on 2018-03-16 15:12:52
 I now see the StreamS app is $5. I won't be paying just to satisfy my curiosity.
We can only make true comparisons when we can encode and decode material of our own choosing.
I hope you're right and xHE-AAC becomes more widely available so that we can evaluate it.

Regarding your listening, it is also worth noting that using wireless headphones can introduce its own complications depending on the codec support of the two devices. Most expensive wireless headphones support AAC. It's possible Opus has to be transcoded to something else before being transmitted over Bluetooth.
Title: Re: xHE-AAC vs HE-AACvE
Post by: imacguru on 2018-03-16 16:01:48
I now see the StreamS app is $5. I won't be paying just to satisfy my curiosity.
We can only make true comparisons when we can encode and decode material of our own choosing.
I hope you're right and xHE-AAC becomes more widely available so that we can evaluate it.

Regarding your listening, it is also worth noting that using wireless headphones can introduce its own complications depending on the codec support of the two devices. Most expensive wireless headphones support AAC. It's possible Opus has to be transcoded to something else before being transmitted over Bluetooth.

Hi Moni!

I've owned many different headphones over the years. The Bose QC35s are the best I've ever had. While they are mostly used via Bluetooth they also can run hard-wired.

I also recall buying numerous "radio" apps until finding StreamS Hi-Fi Radio which clearly had the best sound quality of any in the app store so for me the 5 bucks was money well spent.

I just created a sample file for you by streaming from my iPhone to my Mac using Airfoil Satellite then capturing the result to Apple Lossless Codec with Audio Hijack (using Loopback as a virtual audio device). This should give you a taste of 40k xHE-AAC without buying the app.

This is a link to download the file from my OneDrive: 40kbps xHE-AAC Recorded in Apple Lossless Codec (https://1drv.ms/u/s!AuEfhNk2ZwWBy1-Ms7sCGXxpYOyY)

Note that my streams are in 48kHz sampling rate so beware if what you listen on does any strange sample rate conversions! The file is in Apple Lossless Codec at 48kHz sample rate.

Cheers,
Phil
Title: Re: xHE-AAC vs HE-AACvE
Post by: Moni on 2018-03-16 16:48:34
Thanks for the sample.
As I am currently on a Mac I do not have access to the Opus 1.3 beta but decided to go ahead with a test on 1.2.1.
First off, this is not an apples-to-apples comparison because it is very clear the radio station is processing the audio in a manner that emphasizes the presence region and bass.
Linked are three roughly time-aligned samples. One is the original lossless capture from Spotify (at the "Premium" quality). Another is Opus 1.2.1 encoded at 48kbps. Another is your xHE-AAC. It's very clear which one is your sample because of the aforementioned treble lift. However, they were loudness matched.
I do not notice any major sins of commission in the treble on the Opus encode.
https://ufile.io/spfgw

Unfortunately we can not perform a true comparison until we can encode our own samples.
Title: Re: xHE-AAC vs HE-AACvE
Post by: imacguru on 2018-03-16 17:25:03
Thanks for the sample.
As I am currently on a Mac I do not have access to the Opus 1.3 beta but decided to go ahead with a test on 1.2.1.
First off, this is not an apples-to-apples comparison because it is very clear the radio station is processing the audio in a manner that emphasizes the presence region and bass.
Linked are three roughly time-aligned samples. One is the original lossless capture from Spotify (at the "Premium" quality). Another is Opus 1.2.1 encoded at 48kbps. Another is your xHE-AAC. It's very clear which one is your sample because of the aforementioned treble lift. However, they were loudness matched.
I do not notice any major sins of commission in the treble on the Opus encode.
https://ufile.io/spfgw

Unfortunately we can not perform a true comparison until we can encode our own samples.

Hi Moni,

When time permits, I'll install the Opus 1.3 beta on my stream encoding PC then capture a 40kbps stream off that at the same time as a 40kbps xHE-AAC stream of the same radio source which will be an opportunity for a better comparison.

In the interim, that first sample I provided was done with the iPhone volume at full blast. There is a fair bit of overshoot with playback of anything with SBR so I've dropped the level a touch on the iPhone and grabbed 3 new samples from 3 of my 4 radio stations. One is from the classic rock format, one from the classic soul format and a fresh sample from LG73 when a classic hit aired (Kim Carnes - Bette Davis Eyes, long considered an excellent processor torture test song due to the extreme bass notes that kick in and on many systems cause the vocals to drop way back). The 3 files are combined in a zip archive on my OneDrive at this URL: Three 40kbps xHE-AAC radio stream samples in Apple Lossless Codec at 48kHz Sampling Rate (https://1drv.ms/u/s!AuEfhNk2ZwWBy2CjVTtp5TRD7J1j).

I'll grab your download now and check it out!


Cheers,
Phil
Title: Re: xHE-AAC vs HE-AACvE
Post by: IgorC on 2018-03-17 16:07:45
I’ve lost all interest in xHE-AAC/USAC or in any new MPEG audio standard a long time ago.

Let’s see. MP3, LC-AAC, HE-AACv1 and HE-AACv2 are great. They really are. But then something has happened.
MPEG Surround (standard since 2007) hadn’t see any adoption and now it’s superseded by a new standard (MPEG-H 3D Audio). Just think about it. A whole standard was skipped.  :-\  When people needed most of it during 2007 -2015.

What about xHE-AAC. Is it doing any better than MPEG Surround?  Well, actually it isn’t.
Both xHE-AAC and Opus were released as standards in 2012.
We are seeing  growing support for Opus  https://caniuse.com/#search=opus Right now the index of support is 78%. It will easily hit 90%+ in 1 year.
All members of a “big-four” (Google, Apple, Microsoft and Mozilla) support Opus. And none of them support xHE-AAC. It’s the easiest victory I've ever seen.
And who guarantees that MPEG-H 3D Audio won’t replace xHE-AAC as a superset of it per completely? (as it has happened with MPEG Surround.)

I don’t know what is happening with a new MPEG audio standards during last 10 years but none of them see any meaningful adoption. It’s not my concern or of my interests. Not anymore.

P.S. What in the earth decodes/plays xHE-AAC files if such exists? Leave alone an idea to find an acceptable encoder for that. I won't install VM iOS and pay 5$ for some software which is literally unique program in a whole world which support xHE-AAC. 
Title: Re: xHE-AAC vs HE-AACvE
Post by: saratoga on 2018-03-17 16:54:13
I don’t know what is happening with a new MPEG audio standards during last 10 years but none of them see any meaningful adoption. It’s not my concern or of my interests. Not anymore.


The founder and chairman of MPEG had a recent blog post where he discussed the implosion of MPEG's model:

http://blog.chiariglione.org/2018/01/28/

His conclusion is surprisingly negative about the future of MPEG video standards, but his argument applies even more so to audio. Just as VP9 and it's successors are replacing h.265/hevc, xiph and it's formats will probably displace newer AAC variants in most applications.
Title: Re: xHE-AAC vs HE-AACvE
Post by: IgorC on 2018-03-17 17:39:18
VP9/AV1 make effect on HEVC licensing.

HEVC Advance eliminates streaming royalties https://forum.doom9.org/showthread.php?p=1836640
Title: Re: xHE-AAC vs HE-AACvE
Post by: imacguru on 2018-03-17 21:17:01
I’ve lost all interest in xHE-AAC/USAC or in any new MPEG audio standard a long time ago.

Let’s see. MP3, LC-AAC, HE-AACv1 and HE-AACv2 are great. They really are. But then something has happened.
MPEG Surround (standard since 2007) hadn’t see any adoption and now it’s superseded by a new standard (MPEG-H 3D Audio). Just think about it. A whole standard was skipped.  :-\  When people needed most of it during 2007 -2015.

What about xHE-AAC. Is it doing any better than MPEG Surround?  Well, actually it isn’t.
Both xHE-AAC and Opus were released as standards in 2012.
We are seeing  growing support for Opus  https://caniuse.com/#search=opus Right now the index of support is 78%. It will easily hit 90%+ in 1 year.
All members of a “big-four” (Google, Apple, Microsoft and Mozilla) support Opus. And none of them support xHE-AAC. It’s the easiest victory I've ever seen.
And who guarantees that MPEG-H 3D Audio won’t replace xHE-AAC as a superset of it per completely? (as it has happened with MPEG Surround.)

I don’t know what is happening with a new MPEG audio standards during last 10 years but none of them see any meaningful adoption. It’s not my concern or of my interests. Not anymore.

P.S. What in the earth decodes/plays xHE-AAC files if such exists? Leave alone an idea to find an acceptable encoder for that. I won't install VM iOS and pay 5$ for some software which is literally unique program in a whole world which support xHE-AAC. 


Hi Igor,

AAC still has a rather long list of licensees: AAC Licensees (http://www.via-corp.com/us/en/licensing/aac/licensees.html). Late last year Via Licensing eliminated separate fees for xHE-AAC. That means everyone on that list is already licensed to use xHE-AAC! It takes time for new codecs to gain market share. People still use mp3 even though superior codecs like AAC have existed for a decade and a half! It took several years after HE-AACv2 was introduced till it became widespread with availability at the core operating system level. In the interim many people would listen to HE-AACv2 streams either in wide-band mono (with an HE-AACv1 decoder) or even worse narrow-band mono (AAC-LC decoder).

xHE-AAC has been adopted as the standard for Digital Radio Mondiale. DRM Implementation Guide (http://www.drm.org/DRM_Introduction_and_Implementation_Guide-V3.pdf) All currently shipping DRM receivers have xHE-AAC on board.

The advantages of xHE-AAC include using LOAS instead of ADTS for better sync and less overhead. It works extremely well with HLS for mobile listening.

I think we'd see MPEG-H 3D Audio added to xHE-AAC before we'd see it replace xHE-AAC. xHE-AAC already supports multi-channel surround sound formats.

It's too early to rule out the "big-four" as you refer to them adding and supporting xHE-AAC. I think we'll see that happen later this year for one or two of them.

You've said they all support Opus. I checked Apple for example and do not see Opus anywhere on this page: Supported Audio Codecs in macOS (https://developer.apple.com/library/content/documentation/MusicAudio/Conceptual/CoreAudioOverview/SupportedAudioFormatsMacOSX/SupportedAudioFormatsMacOSX.html)

Another one of the "big-four" also show nothing for Opus: Supported Codecs in Windows (https://docs.microsoft.com/en-us/windows/uwp/audio-video-camera/supported-codecs)

Google are on board with Android 5.0 or newer: Android Supported Codecs (https://developer.android.com/guide/topics/media/media-formats.html) and Mozilla and Chrome browsers support Opus (but other browsers like Edge, Opera and Safari do not). So reality is only two of the "big-four" you refer to support Opus.

All of them support AAC and are now automatically licensed to use xHE-AAC should they decide to do so. How long it might take for that to happen I have no idea but I hope it happens sooner than later. Meanwhile I'll spend some time testing Opus and follow its progress. Always a bonus to have choices!

BTW, to my ears FaceTime Audio blows away Skype audio for quality. So Opus having Skype's SILK codec doesn't guarantee great speech quality. xHE-AAC incorporates AMR-WB+ which many regard as an awesome speech codec.


Cheers,
Phil
Title: Re: xHE-AAC vs HE-AACvE
Post by: saratoga on 2018-03-17 21:39:29

You've said they all support Opus. I checked Apple for example and do not see Opus anywhere on this page: Supported Audio Codecs in macOS (https://developer.apple.com/library/content/documentation/MusicAudio/Conceptual/CoreAudioOverview/SupportedAudioFormatsMacOSX/SupportedAudioFormatsMacOSX.html)

Another one of the "big-four" also show nothing for Opus: Supported Codecs in Windows (https://docs.microsoft.com/en-us/windows/uwp/audio-video-camera/supported-codecs)

Microsoft added Opus support in 2016, Apple in iOS 11.

Since AOM uses Opus for audio, everyone who isn't using HEVC has to implement it, and since HEVC is failing, Opus is becoming universal.  It will be interesting to see what happens with AAC once it is no longer required for video, particularly if they don't cut the license fees.
Title: Re: xHE-AAC vs HE-AACvE
Post by: kode54 on 2018-03-18 07:13:42
macOS High Sierra as well, but only if you pack it in a CAF container instead of Ogg, and the means to do so is undocumented, and the only tool on the OS that supports producing or encoding CAF Opus files only supports mono, and doesn't necessarily guarantee gaplessness.

I produced a patch against FFmpeg that supports encapsulating Opus into CAF containers, but it doesn't generate all of the headers that Apple's tools do, but the files decode nonetheless. The fact that it doesn't produce all the unknown headers, and that said headers are unknown, is why the patch was not accepted.
Title: Re: xHE-AAC vs HE-AACvE
Post by: imacguru on 2018-03-25 22:13:02
Thanks for the sample.
As I am currently on a Mac I do not have access to the Opus 1.3 beta but decided to go ahead with a test on 1.2.1.
First off, this is not an apples-to-apples comparison because it is very clear the radio station is processing the audio in a manner that emphasizes the presence region and bass.
Linked are three roughly time-aligned samples. One is the original lossless capture from Spotify (at the "Premium" quality). Another is Opus 1.2.1 encoded at 48kbps. Another is your xHE-AAC. It's very clear which one is your sample because of the aforementioned treble lift. However, they were loudness matched.
I do not notice any major sins of commission in the treble on the Opus encode.
https://ufile.io/spfgw

Unfortunately we can not perform a true comparison until we can encode our own samples.

Hi Moni,

I checked out the Opus 1.3 beta Windows binaries offered. Then I remembered a much easier way to test Opus: Ladiocast on macOS (current version as of when I'm typing this is 0.12.4)! It has the Opus 1.2.1 release version of the codec. LadioCast is receiving the audio feed from the PC running the Optimod PCn1600 via Rogue Amoeba's Airfoil on the PC and Airfoil Satellite on the Mac (uses Apple AirPlay with Apple Lossless Codec streaming). So no compromise in audio quality getting the feed from the PC to the Mac and into LadioCast! I'm using a virtual audio device created in Loopback to direct the audio out of Airfoil Satellite to Ladiocast. That device is in 2 channel 32 bit floating point mode at a 48kHz sampling rate.

I've setup a test 40kbps average bit-rate 48kHz sampling rate live Opus stream (right now simulcasting MaxRadio.CA) and am listening to it with VLC on my Mac on my AKG K240 Mk II headphones. It sounds very good! I don't think it beats 40kbps xHE-AAC but I would put it in second place over any other 40kbps streaming codecs I've run.

I'm surprised at how good it sounds! Here is the URL of the test stream currently simulcasting my eclectic format station MaxRadio.CA: http://www.maxradio.ca:8000/Opus.m3u (http://www.maxradio.ca:8000/Opus.m3u)

I am hearing the occasional appearance by codec disrupter guy Artie Facts but aside from that Opus is very impressive for an open source codec. For my ears it doesn't currently match the astounding clarity of 40kbps xHE-AAC but maybe it'll catch up in a future version!

Let me know your thoughts on the test stream! I just checked and verified the stream also plays in VLC for iOS. I have not yet tested any other players.


Cheers,
Phil