AVS Forum | Home Theater Discussions And Reviews (https://www.avsforum.com/forum/)
-   2-Channel Audio (https://www.avsforum.com/forum/173-2-channel-audio/)
-   -   Participate in this blind test: Opus vs Vorbis at ~128kbps. Which file sounds better? (https://www.avsforum.com/forum/173-2-channel-audio/3071438-participate-blind-test-opus-vs-vorbis-128kbps-file-sounds-better.html)

Ferrari_1996 06-04-2019 12:59 PM

Participate in this blind test: Opus vs Vorbis at ~128kbps. Which file sounds better?
 
EDIT: I made an online audio converter that allows you to easily convert an audio file to Opus, FLAC, WAV, MP3, AAC or Vorbis. Available at freeaudioconverter.net or onlineaudioconverter.net :)

Methodology:

Original file: Lossless WAV.

I resampled the lossless file to 48kHz using SoX. This is because Opus changes the sample rate to 48kHz and it's best to eliminate extra variables. The resampled version is what I converted to Opus and Vorbis.

For Opus, I used libopus with FFmpeg. VBR mode set to on, bitrate set to 128k.

For Vorbis, I used libvorbis with FFmpeg with the q4 VBR setting as this has a target of ~128k.

After converting the resampled lossless file to Opus and Vorbis, I converted the lossy files to FLAC, so that you cannot check the codecs by downloading the files. I included the codec name in the filenames for my own reference but the uploaded files have been renamed to "1" and "2" as I do not wish to give away which one is which.

You can find the files here: https://drive.google.com/drive/folde...tBkDk1w_z1hSSM

I have included the lossless files too. Both lossless files are labelled.

Artist: CYRYEL

Track name: Entre Toi et Moi (live guitare)

Downloaded from: https://cyryel.bandcamp.com/track/en...i-live-guitare

Which file sounds better to you? "1" or "2"?

stngray72 06-04-2019 01:47 PM

1 st one vocals sound clearer

SeeMoreDigital 06-04-2019 02:38 PM

If you had used say, LameXP to generate your encodes you would have discovered that Opus (Build: 1.3 2018-10-18/24) does not change the sample rate to 48.0KHz. It can generate 44.1KHz encodes perfectly well.

And, I think a VBR encoding bit-rate of 128Kbps is too high for these particular audio formats....

m. zillch 06-04-2019 03:13 PM

Under double blind conditions [Foobar 2000 with the ABX component added] I hear no difference.

Ferrari_1996 06-04-2019 06:36 PM

Quote:

Originally Posted by SeeMoreDigital (Post 58140836)
If you had used say, LameXP to generate your encodes you would have discovered that Opus (Build: 1.3 2018-10-18/24) does not change the sample rate to 48.0KHz. It can generate 44.1KHz encodes perfectly well.

And, I think a VBR encoding bit-rate of 128Kbps is too high for these particular audio formats....

Just tried LameXP and I got a 48kHz sample rate when using the Opus encoder. When I check the properties in foobar, it says the following for the encoder: opusenc from opus-tools 0.2+3-f5f571b3 AVX1 [Oct 24 2018]

I chose 128kbps as that's roughly the bitrate of YouTube audio (opus) and I wanted to see if perhaps they should switch to Vorbis haha.

bing! 06-05-2019 10:46 AM

"In A/V reproduction accuracy, there IS no concept of "accounting for personal taste/preference". As art consumers we don't "pick" the level of bass, nor the tint/brightness of a scene's sky, any more than we pick the ending of a novel or Mona Lisa's type of smile. "High fidelity" means "high truthfulness", faithful to the original artist's intent: an unmodified, neutral, accurate copy of the original master, ideally being exact and with no discernable alterations, aka "transparency"."

Oooooh. We cannot be more different. Museums take great care in the lighting made available to view paintings. Some have sunlights, some have full spectrum lighting, others, not so great. If one cannot see a Dutch realist painting under the sunlight in Holland, you cannot see the painting as the painter saw it. Colors do not pop the same at the equator, in the northern regions et all. At home, if one has the finances to do so, I recommend full spectrum lighting for art viewing. Halogens and standard LED, meh.

No reproduction is absolutely accurate. That is a myth. Most popular modern music and recordings have no real world sound. They are mostly created in the digital domain rendered by whatever the artist has. There is no reproducing the same accurately for all recordings. That endeavor would be costly and fool hardy.

I was a concert tech. Most concerts (The Grateful Dead is one of the exceptions), the artists show up and play music. I have not seen one artist go to stadium seating and have the sound tweaked. It is what it is. Whatever the tech stack plays, thats what it is. Is that what the artist hears on intends to render, no. It is what it is.

In recording, what is heard in the studio is never the same as what's heard in the booth. That is absolute truth. What is heard at home, is even further. Considering a of lot modern recordings are tweaked by the producer to sound how they want it to sound with the most popular playback mediums, the recording itself is not accurate to the artist.

Lastly, my personal tastes in playback were formed from my experiences. I have been in small gatherings with live opera singer and small bands. Garden parties, ballrooms, clubs, home band practices, and large concerts. My memory of what I know how music and instruments should sound like is what dictates how I contour the sound on my rig. If another person wants to go chase charts, thats great. But it is really astounding how such a general statement encompassing the hobby is put out as absolute truth based on a coined term "High Fidelity". On the contrary, high fidelity is a benchmark that should be used as a guide, but still have the flexibility to round off or sharpen the corners. A totally rigid and inflexible attitude to playback sucks the fun out of it. Pass.

m. zillch 06-05-2019 11:16 AM

I'm into high fidelity. [There's no rule everyone has to be.] I'm also willing to bet the OP similarly is seeking out what codecs can make audibly transparent copies of the original file rather than, paraphrased: "Which rendition alters the music in the most pleasing manner according to your ear?"

Ratman 06-05-2019 11:20 AM

Quote:

Originally Posted by bing! (Post 58144252)
"In A/V reproduction accuracy, there IS no concept of "accounting for personal taste/preference". As art consumers we don't "pick" the level of bass, nor the tint/brightness of a scene's sky, any more than we pick the ending of a novel or Mona Lisa's type of smile. "High fidelity" means "high truthfulness", faithful to the original artist's intent: an unmodified, neutral, accurate copy of the original master, ideally being exact and with no discernable alterations, aka "transparency"."

Oooooh. We cannot be more different. Museums take great care in the lighting made available to view paintings. Some have sunlights, some have full spectrum lighting, others, not so great. If one cannot see a Dutch realist painting under the sunlight in Holland, you cannot see the painting as the painter saw it. Colors do not pop the same at the equator, in the northern regions et all. At home, if one has the finances to do so, I recommend full spectrum lighting for art viewing. Halogens and standard LED, meh.

No reproduction is absolutely accurate. That is a myth. Most popular modern music and recordings have no real world sound. They are mostly created in the digital domain rendered by whatever the artist has. There is no reproducing the same accurately for all recordings. That endeavor would be costly and fool hardy.

I was a concert tech. Most concerts (The Grateful Dead is one of the exceptions), the artists show up and play music. I have not seen one artist go to stadium seating and have the sound tweaked. It is what it is. Whatever the tech stack plays, thats what it is. Is that what the artist hears on intends to render, no. It is what it is.

In recording, what is heard in the studio is never the same as what's heard in the booth. That is absolute truth. What is heard at home, is even further. Considering a of lot modern recordings are tweaked by the producer to sound how they want it to sound with the most popular playback mediums, the recording itself is not accurate to the artist.

Lastly, my personal tastes in playback were formed from my experiences. I have been in small gatherings with live opera singer and small bands. Garden parties, ballrooms, clubs, home band practices, and large concerts. My memory of what I know how music and instruments should sound like is what dictates how I contour the sound on my rig. If another person wants to go chase charts, thats great. But it is really astounding how such a general statement encompassing the hobby is put at as absolute truth based on a coined term "High Fidelity". On the contrary, high fidelity is a benchmark that should be used as a guide, but still have the flexibility to round off or sharpen the corners. A totally rigid and inflexible attitude to playback sucks the fun out of it. Pass.

Wonderful.
Did you participate in the test that the OP suggested?

bing! 06-05-2019 11:28 AM

Quote:

Originally Posted by Ratman (Post 58144418)
Wonderful.
Did you participate in the test that the OP suggested?

Nope. My rigs are paralyzed by a short in my half of my house and my garage. Both locations are where my systems are.

I was really interested in doing the tests, but have to wait. An electrician already said he couldnt fix it, I am waiting on an electrical contracting company to come in with test equipment to find the short.

johnplayerson 06-05-2019 11:31 AM

I prefer to leave all the files lossless, The only reason to have a smaller file is space, and possibly planning to play tracks on a source and play medium that will not
reproduce the full sound of the original track anyway. With Flac being approx half the size of wave, and only double the size of a mp3 at 360. I personally see no reason to use a loss driven codec, and see these types of challenges meaningless.

Ratman 06-05-2019 11:52 AM

Quote:

Originally Posted by Ferrari_1996 (Post 58140320)
Which file sounds better to you? "1" or "2"?

Quote:

Originally Posted by bing! (Post 58144462)
Nope. My rigs are paralyzed by a short in my half of my house and my garage. Both locations are where my systems are.

I was really interested in doing the tests, but have to wait. An electrician already said he couldnt fix it, I am waiting on an electrical contracting company to come in with test equipment to find the short.


Super!
Let us know whether "1" or "2" sounds better to you.... when you have your electrical problem(s) resolved. ;)

Ferrari_1996 06-05-2019 02:56 PM

Quote:

Originally Posted by bing! (Post 58144252)
"In A/V reproduction accuracy, there IS no concept of "accounting for personal taste/preference". As art consumers we don't "pick" the level of bass, nor the tint/brightness of a scene's sky, any more than we pick the ending of a novel or Mona Lisa's type of smile. "High fidelity" means "high truthfulness", faithful to the original artist's intent: an unmodified, neutral, accurate copy of the original master, ideally being exact and with no discernable alterations, aka "transparency"."

Oooooh. We cannot be more different. Museums take great care in the lighting made available to view paintings. Some have sunlights, some have full spectrum lighting, others, not so great. If one cannot see a Dutch realist painting under the sunlight in Holland, you cannot see the painting as the painter saw it. Colors do not pop the same at the equator, in the northern regions et all. At home, if one has the finances to do so, I recommend full spectrum lighting for art viewing. Halogens and standard LED, meh.

No reproduction is absolutely accurate. That is a myth. Most popular modern music and recordings have no real world sound. They are mostly created in the digital domain rendered by whatever the artist has. There is no reproducing the same accurately for all recordings. That endeavor would be costly and fool hardy.

I was a concert tech. Most concerts (The Grateful Dead is one of the exceptions), the artists show up and play music. I have not seen one artist go to stadium seating and have the sound tweaked. It is what it is. Whatever the tech stack plays, thats what it is. Is that what the artist hears on intends to render, no. It is what it is.

In recording, what is heard in the studio is never the same as what's heard in the booth. That is absolute truth. What is heard at home, is even further. Considering a of lot modern recordings are tweaked by the producer to sound how they want it to sound with the most popular playback mediums, the recording itself is not accurate to the artist.

Lastly, my personal tastes in playback were formed from my experiences. I have been in small gatherings with live opera singer and small bands. Garden parties, ballrooms, clubs, home band practices, and large concerts. My memory of what I know how music and instruments should sound like is what dictates how I contour the sound on my rig. If another person wants to go chase charts, thats great. But it is really astounding how such a general statement encompassing the hobby is put out as absolute truth based on a coined term "High Fidelity". On the contrary, high fidelity is a benchmark that should be used as a guide, but still have the flexibility to round off or sharpen the corners. A totally rigid and inflexible attitude to playback sucks the fun out of it. Pass.

You're missing the point. The point is to find out which codec is more faithful to the original lossless track (which of course is different to what was heard in the room, but that's beside the point).

Ferrari_1996 06-05-2019 02:58 PM

Quote:

Originally Posted by johnplayerson (Post 58144476)
I prefer to leave all the files lossless, The only reason to have a smaller file is space, and possibly planning to play tracks on a source and play medium that will not
reproduce the full sound of the original track anyway. With Flac being approx half the size of wave, and only double the size of a mp3 at 360. I personally see no reason to use a loss driven codec, and see these types of challenges meaningless.

I prefer leaving files as lossless too. But this is about satisfying my curiosity. I've read multiple times that Opus is the "best" and wanted to put that to the test. Also, those who like to encode to a lossy format to save space may be interested in the result of this experiment (which to be honest won't be very conclusive as it's just one song and a small sample size of participants, but it's something).

bing! 06-05-2019 03:06 PM

Quote:

Originally Posted by Ferrari_1996 (Post 58145486)
You're missing the point. The point is to find out which codec is more faithful to the original lossless track (which of course is different to what was heard in the room, but that's beside the point.

i got that part, I just can't participate right now and caught that other quip.

if I did get to participate, I'd play it through my main 2 channel system and wouldve loved to confirm if I did hear differences and how that compares to others results.

Ferrari_1996 06-05-2019 03:09 PM

Quote:

Originally Posted by m. zillch (Post 58140992)
Under double blind conditions [Foobar 2000 ABX] I hear no difference.

Thanks for participating! I will give it some time before revealing the codecs for the potential of getting more participants that listen without knowing the codecs.

Ferrari_1996 06-05-2019 03:10 PM

Quote:

Originally Posted by Ratman (Post 58144610)
Super!
Let us know whether "1" or "2" sounds better to you.... when you have your electrical problem(s) resolved. ;)

You haven't let us know yourself :p

Ratman 06-05-2019 03:17 PM

Quote:

Originally Posted by Ferrari_1996 (Post 58145540)
You haven't let us know yourself :p

At 128kbps.... I wouldn't bother nor care. :p

Ferrari_1996 06-05-2019 03:21 PM

Quote:

Originally Posted by Ratman (Post 58145568)
At 128kbps.... I wouldn't bother nor care. :p

Too high or too low?

plasma_fan 06-06-2019 11:11 AM

I vote file 1. Listened a few times to 1, 2, and before resample. First place before resample, then 1, last place 2. Did quick switching on my DAC, and my best to match volume.

RayGuy 06-06-2019 05:48 PM

First one sounds thicker or muddier to me, so 2.

Ferrari_1996 06-06-2019 06:58 PM

Quote:

Originally Posted by plasma_fan (Post 58149080)
I vote file 1. Listened a few times to 1, 2, and before resample. First place before resample, then 1, last place 2. Did quick switching on my DAC, and my best to match volume.

Quote:

Originally Posted by RayGuy (Post 58150886)
First one sound thicker or muddier to me, so 2.

Thanks for participating! If either of you want me to reveal the codec of 1 and 2, let me know and I'll send a PM.

plasma_fan 06-06-2019 07:29 PM

Quote:

Originally Posted by Ferrari_1996 (Post 58151122)
Thanks for participating! If either of you want me to reveal the codec of 1 and 2, let me know and I'll send a PM.

Please do send the PM

m. zillch 06-14-2019 07:19 PM

So did you yourself, OP, take the test with Foobar ABX? What were your results?

Ferrari_1996 06-16-2019 06:48 PM

1 Attachment(s)
Quote:

Originally Posted by m. zillch (Post 58183720)
So did you yourself, OP, take the test with Foobar ABX? What were your results?

Nothing jumps out to me when listening to both files via my earphones, so an ABX test would be pointless as I would be guessing. Perhaps 128kbps was too high of a bitrate for a meaningful comparison, as their differences level out at 128kbps, according to the attached image.

NewAger 06-16-2019 11:51 PM

Is Ogg Vorbis still widely used? I remember that being the go-to audio codec for early 2000s video games, a time when developers had to compress the life out of assets to maintain a reasonable storage utilization.

RAART 06-17-2019 07:18 AM

While I had 10 minutes time I listen to original file and the other two...

in first 10 seconds of play, repeated few times, just to make sure... the file number 2

Was listening/using my desktop speakers.

qirex 06-18-2019 01:51 PM

Quote:

Originally Posted by Ferrari_1996 (Post 58190798)
Nothing jumps out to me when listening to both files via my earphones, so an ABX test would be pointless as I would be guessing. Perhaps 128kbps was too high of a bitrate for a meaningful comparison, as their differences level out at 128kbps, according to the attached image.

I'm curious why you're testing at 128 at all, I think even free streaming services are way past that bitrate in 2019.

Ferrari_1996 06-18-2019 03:28 PM

Quote:

Originally Posted by qirex (Post 58198126)
I'm curious why you're testing at 128 at all, I think even free streaming services are way past that bitrate in 2019.

The benefit of Opus is greater at low bitrates. Also, YouTube audio is 128kbps Opus so I wanted to see if perhaps Vorbis would be better.

I'd like to use this opportunity to point out that I don't encode at 128kbps myself. I keep audio files lossless if I have the lossless file.

I'm not promoting the use of a low bitrate. In fact, I recently launched an online audio converter, which allows you to convert to the following formats:

- Opus (libopus 1.3.1)
- Vorbis (libvorbis 1.3.6 with aoTuVb6.03 tune)
- AAC (FDK AAC)
- MP3 (LAME v3.100)
- FLAC
- WAV.

onlineaudioconverter.net

With Opus, Vorbis and AAC, I give the user bitrate options from 32kbps to 480kbps. I coded the backend for the website, so I could have capped the bitrate options at a much lower value. And from what I've read, those codecs reach transparency way before 480. But eh.

The short answer without sneakily advertising my website:

Opus is the king at low bitrates. Also, as you go higher in bitrate, differences between codecs start diminishing, and consequently comparison isn't as important. Heck, according to the image shared in my previous post, it seems like the differences diminish at 128kbps. According to that image, if one is to test whether Opus really is superior, the test should be done at an even lower bitrate.

Ratman 06-18-2019 03:49 PM

Quote:

Originally Posted by Ferrari_1996 (Post 58198682)
The short answer without sneakily advertising my website:

Uh-huh.

m. zillch 06-18-2019 03:50 PM

I like MP3 because of its universality. I've owned products that support MP3, including home/car/portable, but they can't play other formats and there's no way I'm going to spend time converting files. High bitrate MP3 therefor wins in my book [when not using FLAC].


Plus their royalties have expired so there's no concern with encoders or anything either.


All times are GMT -7. The time now is 05:52 AM.

Powered by vBulletin® Copyright ©2000 - 2019, Jelsoft Enterprises Ltd.
vBulletin Security provided by vBSecurity (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
User Alert System provided by Advanced User Tagging (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.