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: FLAC VS OPUS @ 170 Kbps (Read 22048 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

FLAC VS OPUS @ 170 Kbps

So I do some abx in my free time to find setting setting that was transparent to me and I was quite surprise by the result of the test


foo_abx 2.0.1 report
foobar2000 v1.3.9
2015-11-18 17:09:33

File A: 01 - Awake.flac
SHA1: 8978e4aa16dab9a0f5ad59f61b7295b22c85415e
File B: 01 - Awake.opus
SHA1: fcd45fd5639b2601708c8193bfb76067baa1ae99

Output:
DS : Primary Sound Driver
Crossfading: NO

17:09:33 : Test started.
17:09:59 : 01/01
17:10:34 : 02/02
17:13:26 : 03/03
17:14:25 : 03/04
17:14:53 : 04/05
17:15:37 : 04/06
17:16:21 : 05/07
17:16:51 : 06/08
17:17:52 : 07/09
17:18:30 : 08/10
17:19:31 : 09/11
17:20:07 : 10/12
17:20:32 : 11/13
17:21:06 : 11/14
17:21:21 : 12/15
17:21:40 : 13/16
17:21:40 : Test finished.

----------
Total: 13/16
Probability that you were guessing: 1.1%

-- signature --
05120a87703e514dc401be4513c2ee7bdc612102

I also do abx with FLAC VS OPUS @ 127 Kbps and I get the same result but I don't save that log since I though everybody can totally hear 127 kbps and lossless. So what you guys think?
I will think about tomorrow's problem tomorrow

FLAC VS OPUS @ 170 Kbps

Reply #1 –
I think it's a shame Peter is so busy with mobile foobar2000 development that foo_abx is still broken. You can't ABX files with different samplerates because the component reveals which one is playing by accident.

If you wish to do a fair comparison you need to go to player preferences and activate one of the resampler DSPs and set it to resample everything to 48 kHz. Then when you start the ABX test enable the use of current DSP settings.

Also 128 kbps Opus is definitely not automatically noticeable from the original.

FLAC VS OPUS @ 170 Kbps

Reply #2 –
Quote from: Case on
I think it's a shame Peter is so busy with mobile foobar2000 development that foo_abx is still broken. You can't ABX files with different samplerates because the component reveals which one is playing by accident.

If you wish to do a fair comparison you need to go to player preferences and activate one of the resampler DSPs and set it to resample everything to 48 kHz. Then when you start the ABX test enable the use of current DSP settings.

Also 128 kbps Opus is definitely not automatically noticeable from the original.

No its not different sample rate. I resample flac to 48kHz. If don't resample I will get a perfect result with 0% probability of guessing since there is a pause with the file that has different sample rate. So I will detect it immediately and I won't need 11 minute to identify all of those track
I will think about tomorrow's problem tomorrow

FLAC VS OPUS @ 170 Kbps

Reply #3 –
Did you resample the FLAC first to 48000 Hz and then created the Opus from this resampled file? Otherwise you maybe used two different resamplers.

But isn't that normal anyway? I thought the the bitrate for transparency was a little bit higher.
Test for transperency

Will test myself after work. Bye

FLAC VS OPUS @ 170 Kbps

Reply #4 –
How do you use the gap to detect which is which? Doesn't it just tell you when there is a switch, not what the switch is to?

FLAC VS OPUS @ 170 Kbps

Reply #5 –
Quote from: dhromed on
How do you use the gap to detect which is which? Doesn't it just tell you when there is a switch, not what the switch is to?

I can't understand what your point. Sorry my English isn't that good. How about you try abx 2 file with different sample rate you will notice it right away based on the gap.

Edit : I can understand you now. File with the same sample rate doesn't have the gap between them so if you press X and then press A and you notice there's a huge gap. Then you press X and the press B, there you notice there is only a little gap. From that you already know that X is B and not A based on the sample rate
I will think about tomorrow's problem tomorrow

FLAC VS OPUS @ 170 Kbps

Reply #6 –
Quote from: rudolf.gelpke on
Did you resample the FLAC first to 48000 Hz and then created the Opus from this resampled file? Otherwise you maybe used two different resamplers.

But isn't that normal anyway? I thought the the bitrate for transparency was a little bit higher.
Test for transperency

Will test myself after work. Bye

Yup I resampled flac first then convert it to opus. I also do abx on qaac @ 192 kbps and I success on the test. So I guess 192 kbps aac is still not transparent for me. I still has the log if you want it
I will think about tomorrow's problem tomorrow

FLAC VS OPUS @ 170 Kbps

Reply #7 –
Quote
huge gap | little gap


ahhh thanks.
I thought the gap was always the same.

FLAC VS OPUS @ 170 Kbps

Reply #8 –
Quote
Yup I resampled flac first then convert it to opus. I also do abx on qaac @ 192 kbps and I success on the test. So I guess 192 kbps aac is still not transparent for me. I still has the log if you want it


Not necessary with the logs. Just wanted to make sure you didn't make the same resample mistake like i did a week ago when i played around with Opus. ^^
On the other hand i believe that you hear a difference and it's not transparent for you. But this of course depends on the songs.
Very often there are just specific spots in a song where you can hear a difference. So you have to know what to listen for.

Which song did you test by the way?

FLAC VS OPUS @ 170 Kbps

Reply #9 –
Quote from: rudolf.gelpke on
Quote
Yup I resampled flac first then convert it to opus. I also do abx on qaac @ 192 kbps and I success on the test. So I guess 192 kbps aac is still not transparent for me. I still has the log if you want it


Not necessary with the logs. Just wanted to make sure you didn't make the same resample mistake like i did a week ago when i played around with Opus. ^^
On the other hand i believe that you hear a difference and it's not transparent for you. But this of course depends on the songs.
Very often there are just specific spots in a song where you can hear a difference. So you have to know what to listen for.

Which song did you test by the way?

Its a Japanese song called Awake sing by yuiko. If you want it I can upload it somewhere. The song is quite complex.
I will think about tomorrow's problem tomorrow

FLAC VS OPUS @ 170 Kbps

Reply #10 –
Quote from: rudolf.gelpke on
But isn't that normal anyway? I thought the the bitrate for transparency was a little bit higher.
Test for transperency

That's obviously an individual thing; for me it ranges from 90-150kbps depending on the content, mostly on the low end. The only things I can ABX above 130 or so are a few songs with echoing sibilant whispers overlaid on a busy track, and they already sound distorted, so I'm not surprised. I'm shocked anyone can partially ABX 170 at all, I could never, but good ears and good equipment means something.

FLAC VS OPUS @ 170 Kbps

Reply #11 –
Found my sweet spot. This song Awake by yuiko is really my sweet spot. I just do another abx with the original file (without resample) and mp3 V1 and this is the result

foo_abx 2.0.1 report
foobar2000 v1.3.9
2015-11-19 18:14:11

File A: 01. yuiko — Awake.flac
SHA1: 35140ac2c8edc74a8681be43878c30abf6d700ec
File B: 01 - Awake.mp3
SHA1: bbbee4689d0561bfba757e30df19e35f7d22eadc

Output:
DS : Primary Sound Driver
Crossfading: NO

18:14:11 : Test started.
18:15:14 : 01/01
18:15:45 : 02/02
18:16:42 : 03/03
18:17:22 : 04/04
18:18:00 : 04/05
18:18:36 : 04/06
18:19:41 : 05/07
18:20:29 : 06/08
18:21:04 : 07/09
18:21:41 : 08/10
18:22:15 : 08/11
18:22:51 : 09/12
18:23:29 : 09/13
18:24:18 : 10/14
18:26:07 : 11/15
18:27:01 : 12/16
18:27:01 : Test finished.

----------
Total: 12/16
Probability that you were guessing: 3.8%

-- signature --
f53dc1d4f1f96014fc2a0623f9235cb091ef82d6
I will think about tomorrow's problem tomorrow

FLAC VS OPUS @ 170 Kbps

Reply #12 –
For 170 ~ 192k Try Musepack  Q5 or 6. It could have a slight advantage over the others in some aspects. Also 1 track doesn't always represent the majority. Also you might just accept not fully perfect but not annoying level.

FLAC VS OPUS @ 170 Kbps

Reply #13 –
Quote from: shadowking on
For 170 ~ 192k Try Musepack  Q5 or 6. It could have a slight advantage over the others in some aspects. Also 1 track doesn't always represent the majority. Also you might just accept not fully perfect but not annoying level.

I would like to but android doesn't support musepack. I use flac for my home listening through my PC. I'm searching for ideal codec that can be used for my android. I don't have time to test all 900 track of my music so I will just stick with 1 or 2 track. I will test with another track later. The major aspect that I hear between mp3 and flac is mp3 has a less powerful bass than its lossless version and it not as loud as it use to be on flac but only for a bit
I will think about tomorrow's problem tomorrow

FLAC VS OPUS @ 170 Kbps

Reply #15 –
Quote from: j8ee on
Quote from: Incuriousity on
/.../
android doesn't support musepack.
/.../


https://www.musepack.net/index.php?pg=pro - seem to be a lot of players that does?

I only want player that support embedded lyric and a good ui. I already try Rocketplayer but I need to pay for it to play musepack. I will just wait for android 6.0 and encode all of my song to opus. Right now AAC 224 kbps seems good
I will think about tomorrow's problem tomorrow

FLAC VS OPUS @ 170 Kbps

Reply #16 –
Also although 13/16 is a good valid result it is not perfect or as good at least 14/16.  You want pval < 1 %  unless you really hear something annoying.  12/16 is valid but not too great. I suggest many trials esp when the differences aren't too obvious like 35 ~ 50 trials .  Repeated a positive abx is important when you have weird things like this sample. Looking at the logs it doesn't look like the problem is easy to spot at those settings.

FLAC VS OPUS @ 170 Kbps

Reply #17 –
Quote from: Incuriousity on
Its a Japanese song called Awake sing by yuiko. If you want it I can upload it somewhere. The song is quite complex.

You could upload it to Mega so it is encrypted and send me the PM so you won't get problems with copyright. Or just cut it to 30-60 seconds.

Quote from: foxyshadis on
Quote from: rudolf.gelpke on
But isn't that normal anyway? I thought the the bitrate for transparency was a little bit higher.
Test for transperency

That's obviously an individual thing; for me it ranges from 90-150kbps depending on the content, mostly on the low end. The only things I can ABX above 130 or so are a few songs with echoing sibilant whispers overlaid on a busy track, and they already sound distorted, so I'm not surprised. I'm shocked anyone can partially ABX 170 at all, I could never, but good ears and good equipment means something.

My range is similar to yours with how good the encoders got now. Even if i could ABX it, most of the time it wouldn't bother me during normal listening.
There is only one case where it does bother, but i don't want to talk about it here.
Friend of me has better equipment and i can ABX a higher bitrate than at home. But even killer samples you can find here on this forum sound fine now above 160 - 192 kbps AAC (or even lower).
So i am just interested if i will be able to ABX it.

Probably Offtopic:
It's often interesting to let friends who say that all lossy codecs sucks do an actual test and they start sweating when they can't ABX files at 128 kbps already.
And they tell you before that 320 kbps MP3 are barely passable. And even after that their mind still don't trust the encoder.
Or trying to explain why VBR should be better than CBR. "But what if the encoder decides wrong? How can he know when it is enough?".
But i must say they are more informed now. Still using 256 CVBR AAC on the iPod since we have enough space and just to ease their minds. :-D

My cousin still don't like me cause he thinks ABX is not necessary and just testing the two files with VLC is enough.
So in his eyes i am calling him a liar, because i think thats not enough for a test and don't really believe that he can easily find the high bitrate MP3 files.

FLAC VS OPUS @ 170 Kbps

Reply #18 –
Quote from: shadowking on
Also although 13/16 is a good valid result it is not perfect or as good at least 14/16.  You want pval < 1 %  unless you really hear something annoying.  12/16 is valid but not too great. I suggest many trials esp when the differences aren't too obvious like 35 ~ 50 trials .  Repeated a positive abx is important when you have weird things like this sample. Looking at the logs it doesn't look like the problem is easy to spot at those settings.

35 - 50 trial is too much for me. I hope this log will satisfied you.

FLAC VS OPUS @ 170 Kbps

foo_abx 2.0.1 report
foobar2000 v1.3.9
2015-11-20 17:03:54

File A: 01. yuiko — Awake.flac
SHA1: 35140ac2c8edc74a8681be43878c30abf6d700ec
File B: 01 - Awake.opus
SHA1: c404ed81492e61e070f1beb2afa8ee3eee5929ac

Used DSPs:
Equalizer, Resampler (dBpoweramp/SSRC)

Output:
DS : Primary Sound Driver
Crossfading: NO

17:03:54 : Test started.
17:04:48 : 01/01
17:05:25 : 02/02
17:06:10 : 03/03
17:06:46 : 04/04
17:07:25 : 05/05
17:08:19 : 06/06
17:08:59 : 07/07
17:09:35 : 08/08
17:10:10 : 08/09
17:10:46 : 09/10
17:11:23 : 10/11
17:12:01 : 11/12
17:12:35 : 12/13
17:13:13 : 13/14
17:13:29 : 14/15
17:14:05 : 15/16
17:14:40 : 16/17
17:15:18 : 17/18
17:15:33 : 18/19
17:16:10 : 19/20
17:16:10 : Test finished.

----------
Total: 19/20
Probability that you were guessing: 0.0%

-- signature --
88248f1d5a9f1c9888f8f45d84a5c60164ae43b7

Rudolf.gelpe I will upload it on Sunday cause I don't have access to internet on my computer right now.
I will think about tomorrow's problem tomorrow

FLAC VS OPUS @ 170 Kbps

Reply #19 –
Quote from: Incuriousity on
Rudolf.gelpe I will upload it on Sunday cause I don't have access to internet on my computer right now.


The Opus developers are probably quite interested in it.

FLAC VS OPUS @ 170 Kbps

Reply #21 –
Might be I miss something but what OPUS version was tested? 1.1?

FLAC VS OPUS @ 170 Kbps

Reply #22 –
Quote from: alter4 on
Might be I miss something but what OPUS version was tested? 1.1?

Yes 1.1. It was the latest stable version am I right?

Latest test

OPUS @ 199 kbps (I set it to 200 but I get 199)

foo_abx 2.0.1 report
foobar2000 v1.3.9
2015-11-21 21:14:54

File A: 01. yuiko — Awake.flac
SHA1: 35140ac2c8edc74a8681be43878c30abf6d700ec
File B: 01 - Awake.opus
SHA1: e44768917ad47640522ac96b61bd218b88a99af7

Used DSPs:
Resampler (dBpoweramp/SSRC), Equalizer

Output:
DS : Primary Sound Driver
Crossfading: NO

21:14:54 : Test started.
21:15:59 : 00/01
21:16:44 : 00/02
21:17:50 : 01/03
21:18:47 : 02/04
21:19:25 : 03/05
21:20:02 : 04/06
21:20:38 : 05/07
21:21:14 : 06/08
21:22:11 : 07/09
21:23:13 : 08/10
21:23:50 : 08/11
21:24:45 : 09/12
21:25:38 : 10/13
21:26:15 : 11/14
21:26:53 : 11/15
21:27:29 : 12/16
21:28:22 : 13/17
21:29:21 : 14/18
21:30:00 : 15/19
21:30:56 : 16/20
21:30:56 : Test finished.

----------
Total: 16/20
Probability that you were guessing: 0.6%

-- signature --
74858496726deef77e7f720ead5c1607d61d370aa

It sound better than 170 kbps but the different with FLAC is still audible
I will think about tomorrow's problem tomorrow

FLAC VS OPUS @ 170 Kbps

Reply #23 –
@Incuriousity
Would be interesting to see if it would help when you raise the framesize to 60 ms with Opus.
Since you don't need the latency anyway for music listening in your case.
Helped me in some tests. But might not help much with higher bitrates.

Code: [Select]
--framesize 60

FLAC VS OPUS @ 170 Kbps

Reply #24 –
Can you hear a difference without the Equalizer?