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: Opus 1.3-beta is here (Read 72718 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Re: Opus 1.3-beta is here

Reply #150
Leo '69'

What part of TOS5 You don't understand
Quote
Submission of a thread that is not related to any of the topics covered by the Hydrogenaudio forums and subforums, or posts which are not related to the thread that they are submitted to, is against the rules. If you wish to discuss Off-Topic issues, please use our Off-Topic forum. Repeated violation of this rule after one notification may lead to administrative action including banishment.
So not even being a moderator here, You decided to publicly charge me with some imaginary violations? While I DO respect your reputation on these forums and couple of others, you should stop going overboard.

You'd better answer the guy, who sincerely wants to help improve the already great codec and devote a lot of time to this - instead of being a douche. Sorry for being straightforward, but that's me.

@IgorC, while not obvious at first sight (I was also puzzled originally), I believe Leo 69's post was (sufficiently) on topic. I believe he was asking about audio gear because juliobbv looked like he wanted to do some listening tests. As for his reply to punkrockdude it seems to be related to an earlier comment in the thread about AAC vs Opus comparison.

@Leo 69, please no need for personal insults here. Your post looked a bit off topic at first sight and I believe IgorC was acting in good faith, even if a little harsh.

Now anyone has any comments on the latest Opus builds?  ;)

Re: Opus 1.3-beta is here

Reply #151
jmvalin,

No, no, no
 He wasn't on topic here.  Leo just asked to test AAC absolutely nothing to do with "Opus vs AAC" . I followed topic.

He was off-topic 100%.

So not even being a moderator here, You decided to publicly charge me with some imaginary violations? While I DO respect your reputation on these forums and couple of others, you should stop going overboard.
And yes, I will indicate you to stop off-topic.
I'm a member of this forum and I have right to indicate to You to stop off-topic. You're violating rules of this forum.


You'd better answer the guy, who sincerely wants to help improve the already great codec and devote a lot of time to this. Instead of being a douche. Sorry for being so straightforward, but that's me.
Irony.  :)   Do you understand that I was around and involved all this time in every freaking Opus version, testing it, organinizing tests etc.....
I don't remeber You being involved .... again, Who are You?  And what makes you think You can tell me "You'd better answer" ..... F.O.

Do You understand that my collaboration with Jean-Marc has moved to private mail because absolutely noone here have submit one single meaninful result. This is thread is full nonsense and off-topic. 

The only guys who here "sincerely want" (as you said) to help improve the codec is me and JM himself. I'm not trying to be a superman but  there isn't much help really. We don't see much  help from You or any other person.

If You're not helping at all.... at least stop off-topic.

Re: Opus 1.3-beta is here

Reply #152
@jmvalin
Thank you very much for your comments. They are 100% on point. Someone just can't read between the lines..my goodness.


Re: Opus 1.3-beta is here

Reply #153
As somebody who has never done any ABC/HR or ABX tests (ironically!): can you help me with recommended resources on how to set up everything, in terms of installing software and preparing the encoded samples for the listening tests? I use Windows 10. Also, let me know if there are any samples you want me to focus on first, and which Opus encoders should I use to encode and compare the samples. I'll give it a shot over the long weekend.
Here are some hopefully useful links https://hydrogenaud.io/index.php/topic,115156.msg952288.html#msg952288

Also You can take a look here https://hydrogenaud.io/index.php/topic,16295.0.html

Re: Opus 1.3-beta is here

Reply #154
Hi Julio! And, welcome to the club. What's your audio gear?

Hi Leo. Thanks for the welcome! As for my audio gear, I have these Audio-Technica ATH-M30x headphones: https://www.amazon.com/gp/product/B00HVLUQW8.

I read ff123's "Practice With ABC/HR" guide (https://web.archive.org/web/20110611013208/http://ff123.net/64test/practice.html) to get ABC-HR set up on my machine, and went over the suggested "newkid" session. After that, I used one of @IgorC's samples to create my own "smoke-test" session to verify that I did set up everything right.

Code: [Select]
1L = smoketest\02 fatboy_30sec Opus 1.2.1 32 kbps.wav
2R = smoketest\02 fatboy_30sec Opus 1.3b1 32 kbps.wav

---------------------------------------
1L File: smoketest\02 fatboy_30sec Opus 1.2.1 32 kbps.wav
1L Rating: 3.0
1L Comment: Very odd voice change at 1s. 'Ringing' artifacts appear to be sligtly worse than sample 2.
---------------------------------------
2R File: smoketest\02 fatboy_30sec Opus 1.3b1 32 kbps.wav
2R Rating: 3.8
2R Comment: First part (with the solo robotic voice) almost perfect. Second part contains some 'ringing' artifacts (esp. at 16.5s).
---------------------------------------

ABX Results:


It looks like I should be good by now -- I'm ready to try out more samples. What interesting samples/encoder combinations would you like for me to try out? I'm open on ideas.

Edit:

Here are some hopefully useful links https://hydrogenaud.io/index.php/topic,115156.msg952288.html#msg952288

Also You can take a look here https://hydrogenaud.io/index.php/topic,16295.0.html

Thanks for the links -- I'll go take a look at them.

Re: Opus 1.3-beta is here

Reply #155
You can try any of your own samples  or try from these sources

http://listening-tests.hydrogenaud.io/igorc/aac-96-a/all_samples.zip
http://www.rarewares.org/test_samples/
http://www.mp3-tech.org/tests/aac_48/samples/
http://listening-test.coresv.net/results.htm


At this point it's worth to try these builds vs stable 1.2.1 at 32-48 kbps.
https://hydrogenaud.io/index.php/topic,115156.msg952996.html#msg952996

Re: Opus 1.3-beta is here

Reply #156
I have also been following Opus development for a long time but have not participated in any sort of listening test I could submit to JMV for improvement of the codec.
I am the director of concert recordings at a conservatory and we are currently in the initial planning stages for what our next digital platform will be to make these recordings available. I would very much like for Opus to be one of the flavors available. I realize I have an opportunity with the large body of recordings I work with, and my calibrated office with good equipment, to contribute. Classical and opera can be difficult to encode during quiet or solo portions, as they lack the large masking effect provided by most modern music recordings, and the importance of preserving the timbre.
Anyhow, thanks to the initial question from @juliobbv and helpful response from @IgorC , I will start some testing in the next few weeks.

 

Re: Opus 1.3-beta is here

Reply #157
I read ff123's "Practice With ABC/HR" guide (https://web.archive.org/web/20110611013208/http://ff123.net/64test/practice.html) to get ABC-HR set up on my machine, and went over the suggested "newkid" session. After that, I used one of @IgorC's samples to create my own "smoke-test" session to verify that I did set up everything right.

Code: [Select]
1L = smoketest\02 fatboy_30sec Opus 1.2.1 32 kbps.wav
2R = smoketest\02 fatboy_30sec Opus 1.3b1 32 kbps.wav

...

Just for the record, your test is  in-line with my results. Good.
At this point it would be great to try speech or mixed speed/music samples at 16-32-40 kbps. https://hydrogenaud.io/index.php/topic,115156.msg952996.html#msg952996

I don't want to spoil a surprise (this is a phrase when actually a surprise will be spoiled   :)  ) but  'speech/music classification using an RNN' starts to work after ironing out issues.
At least for speech, Opus starts to wipe the floor with a modern codecs like xHE-AAC/USAC, EVS. (leave alone outdated HE-AAC v1/v2 at this point) . Well, all modern codecs are already on par on music at 48 kbps and higher.
Something big is coming.


Re: Opus 1.3-beta is here

Reply #158
Do You understand that my collaboration with Jean-Marc has moved to private mail because absolutely noone here have submit one single meaninful result.

That's really pity, I was really interrested in reading about the progress of Opus and the tests - unfortunatelly I cannot contribute just now as I didn't find any suitable programm for Linux to do the ABX and I don't know if it would make sense with my current equipment (chep laptop speakers)...

Re: Opus 1.3-beta is here

Reply #159
Do You understand that my collaboration with Jean-Marc has moved to private mail because absolutely noone here have submit one single meaninful result.

That's really pity, I was really interrested in reading about the progress of Opus and the tests - unfortunatelly I cannot contribute just now as I didn't find any suitable programm for Linux to do the ABX and I don't know if it would make sense with my current equipment (chep laptop speakers)...

For Linux, use squishyball, which is the Xiph.org ABX tool.

For Ubuntu and Debian, a "sudo apt install squishyball" should do the job.

Of course, laptop speakers are most likely not useful for the job... but even cheap headphones can yield interesting results.

Re: Opus 1.3-beta is here

Reply #160
After the inital test in abc-hr, I was dismayed to see how much work goes into creating one's own tests. I found abc-hr for Java here (http://www.rarewares.org/others.php) due to a link in this document: www.rarewares.org/rja/ListeningTest.pdf
I just wanted to check that results from this program are considered valid as well.

I am going to start assembling some of my own samples from audiobooks and lectures on music (alternating speech and music) and see what I get, using the builds referenced by IgorC.

Re: Opus 1.3-beta is here

Reply #161
I  think that opus new tf 2 sounds better at 24 kbps than newtf.
64 kbps is similar ( not so big improvement,  maybe it's better in some samples.)
NEW TF 2 it's not that bad.


Re: Opus 1.3-beta is here

Reply #163
This beta release of the upcoming Opus 1.3 includes:
  • Enabling by default the spec fixes in RFC 8251
  • Improvements to the VAD and speech/music classification using an RNN
  • Improvements to stereo speech coding at low bitrate
  • Added support for ambisonics projection using mapping 3 (disabled by default)
  • Fixes to the CELT PLC
Additionally, as a way to test the upcoming update to opus-tools, we’re providing Windows binaries built with 1.3-beta. These binaries are based on libopusenc, which means opusenc is finally able to make use of the Opus delayed-decision feature to make better speech/music transitions.

Source code: opus-1.3-beta.tar.gz
Win32 binaries (experimental): opus-tools-test1.zip

Moin,
What's happen if add two-pass option?

Re: Opus 1.3-beta is here

Reply #164
Suggested patch to enable progress output (percentage) in "opusdec" tool (new libopusfile-based version):
https://gist.github.com/anonymous/2ea78e68a251f7fdb6e46822575af2eb

This also limits the update frequency a bit, because current version writes updates very frequently, which can be slow on Windows console...

Small update with some clean-up:
https://gist.github.com/anonymous/103196c96a7f795c37f2cabffe7dbf03

Re: Opus 1.3-beta is here

Reply #165
What's happen if add two-pass option?
If you're referring to the delayed-decision option, it's not two-pass, it's really just a 2-second look-ahead in the analysis that makes it possible to make better speech/music decisions. It also helps a little with tone onsets. The advantage of this being a delay and not a two-pass option is that it can still be used for real-time streaming if a short delay is acceptable (it can also be configured to be less than 2 seconds). Of course, for VoIP and videoconferencing there's no delayed decision at all since any delay is bad.

Re: Opus 1.3-beta is here

Reply #166
Found a bug with v1.3beta. High frequencies suffer for a noticeable period, compared to v1.2.
Please download: http://wdfiles.ru/5lXj (big blue single button in the middle of the page).
Use the phrase "haopusbeta" to open.
Apologies for full composition, couldn't find a way to show the bug other way.

Re: Opus 1.3-beta is here

Reply #167
Is there an update to OPUS beta?  :)

Re: Opus 1.3-beta is here

Reply #168
You already have a scale, it's called "bitrate". Instead of having a 0-10 scale that maps to different bitrates (because Vorbis also calibrates its quality scale on an average bitrate), the scale is 0-510 you don't even have to look up what value you need to get a certain average bitrate.
I know I'm VERY late with the answer, but I completely forgot I asked this question.

I don't have a problem with this approach when it comes to stereo, the problem is achieving the same quality in multichannel files. Currently I have pretty good heaphones and a mediocre 5.1 setup, and I would like to encode a futureproof 5.1 file (in case I update my setup to something better) - with the same quality level that I was able to determine is transparent on headphones. There is really no other way to do that than to guess what bitrate will be appropriate, or take ABX tests again - which I don't want to do, because, as I said, my setup isn't the greatest.

With qaac it's easy - I know that TVBR 109 is transparent for me, so I can just encode any file at this setting, no matter how many channels it has, and I will always end up with bitrate that's appropriate for the encoded content.