IPB

Welcome Guest ( Log In | Register )

3 Pages V   1 2 3 >  
Reply to this topicStart new topic
How is the quality of MusePack SV8, vs. newer codecs like AAC, Vorbis?, [was “Question about musepack sv8”/ToS #6]
B7k
post Aug 28 2013, 03:36
Post #1





Group: Members
Posts: 45
Joined: 11-March 07
From: cleveland,ohio
Member No.: 41371



Hi It's been sometime since I used this codec I have a several old albums encoded with musepack. I use an Ipod video 80gb with rockbox and have since started using autov vorbis and wondered how sv8 compares with todays modern codecs such as apple/nero aac or vorbis.
Go to the top of the page
+Quote Post
shadowking
post Aug 28 2013, 05:37
Post #2





Group: Members
Posts: 1526
Joined: 31-January 04
Member No.: 11664



Depends at the bitrate your targeting . Below 128k no question AAC / vorbis are superior. The real strength of MPC is the default profile @ 160..180k yields very high quality. I would trust it over modern 256~320k encodings of mp3 / acc . Vorbis aotuv could be same quality maybe even better but no one knows for sure. I still think MPC has an advantage in some areas like pre-echo. Battery life should be better with MPC vs vorbis.

I also like that MPC has all the tools and encoders developed in one place.


--------------------
Wavpack -b450s0.7
Go to the top of the page
+Quote Post
eahm
post Aug 28 2013, 06:12
Post #3





Group: Members
Posts: 1080
Joined: 11-February 12
Member No.: 97076



shadowking, you would put MPC at ~160-180 kbps over ~256-320 kbps AAC? Are you also saying that Ogg Vorbis at ~160-180 is "better" than AAC at ~256-320?

This post has been edited by eahm: Aug 28 2013, 06:18


--------------------
/lwAsIimz
Go to the top of the page
+Quote Post
shadowking
post Aug 28 2013, 09:38
Post #4





Group: Members
Posts: 1526
Joined: 31-January 04
Member No.: 11664



No , I think vorbis 'better' or on-par may be at higher bitrates like Q7 (224k) . Q6 at the very least (190k).. I still think at moderate-high bitrate (160..210k) mpc @ Q5 / 6 has the edge over AAC 256 and mp3 320 and even vorbis. For an 'extreme' setting i'd take mpc Q6 (205k) anyday over the others @ 256-320. Even mpc 'insane' profile yields lower bitrate (240k) while allowing full bandwidth encoding which the others don't. Even at medium bitrates say 130..160k mpc (--radio / Q4 ~4.5) will perform very well and is well suited for portable use. Its just that it will pretty much suck at 100k or lower and vorbis / aac are better 'overall' performers in that sense.

Another observation is that mpc is a really fast encoder without any 3rd party versions. Faster on my pc than lame mp3, aac, vorbis and close to aotuv lancer.

This post has been edited by shadowking: Aug 28 2013, 09:48


--------------------
Wavpack -b450s0.7
Go to the top of the page
+Quote Post
larryfine
post Aug 28 2013, 19:27
Post #5





Group: Members
Posts: 57
Joined: 21-September 10
Member No.: 84040



QUOTE
Battery life should be better with MPC vs vorbis.


I fully agree.
In my testing with MPC SV8 on android device realized a reduction of approximately 2~3% in CPU usage and, as a result, longer battery life.


--------------------
loquor mee menti: factus de materia, cinis elementi...
Go to the top of the page
+Quote Post
darkbyte
post Aug 28 2013, 19:32
Post #6





Group: Members
Posts: 152
Joined: 14-June 11
Member No.: 91517



I wonder how Musepack's subbands align with SBC's subbands. I'm currently using LossyFlac extraportable to provide as slightly modified audio file as possible to SBC encoding to achieve better quality but maybe because mpc is based on mp2 and SBC is based on mp1 it might be a good choice aswell with half the bitrate.

In my tests MPC is very competetive in quality at Q5 and encoding and decoding speed is amazingly fast. Although i've just used it seldomly because other formats are more common if i have to share them with somebody else.

This post has been edited by darkbyte: Aug 28 2013, 19:34


--------------------
Wavpack -b450x1c
Go to the top of the page
+Quote Post
eahm
post Aug 28 2013, 21:53
Post #7





Group: Members
Posts: 1080
Joined: 11-February 12
Member No.: 97076



shadowking, you talk about MPC like it's the best lossy codec ever created. I understand it may be good but I wouldn't even compare it to AAC, not even once.

Let me understand one thing, if it's so good why did they stop adding it into listening test since 2004? It's obsolete.


--------------------
/lwAsIimz
Go to the top of the page
+Quote Post
saratoga
post Aug 28 2013, 22:04
Post #8





Group: Members
Posts: 4959
Joined: 2-September 02
Member No.: 3264



QUOTE (eahm @ Aug 28 2013, 16:53) *
Let me understand one thing, if it's so good why did they stop adding it into listening test since 2004?


Because it was found to be transparent over the range of bitrates it was made for. No sense doing listening tests if they can't tell you anything new.
Go to the top of the page
+Quote Post
B7k
post Aug 28 2013, 22:33
Post #9





Group: Members
Posts: 45
Joined: 11-March 07
From: cleveland,ohio
Member No.: 41371



So it looks like the codec is on par with today's modern codecs because of everything done to it and is finalized. Thats the reason I could not find an updated listening test to compare it with.
Go to the top of the page
+Quote Post
Canar
post Aug 28 2013, 23:58
Post #10





Group: Super Moderator
Posts: 3361
Joined: 26-July 02
From: princegeorge.ca
Member No.: 2796



Terms of Service 8 called. Apparently no one here knows it so they never picked up... >_>


--------------------
You cannot ABX the rustling of jimmies.
No mouse? No problem.
Go to the top of the page
+Quote Post
B7k
post Aug 29 2013, 01:50
Post #11





Group: Members
Posts: 45
Joined: 11-March 07
From: cleveland,ohio
Member No.: 41371



Well I encoded 2 albums different genres and Indeed their is better battery life with rockbox compared with the same albums encoded in vorbis. I find musepack indistinguishable from the flac files i encoded from. Thanks for the support everyone who answered this thread but I have the feeling some of us might have got off topic and violated Tos 8 of the forums.
Go to the top of the page
+Quote Post
eahm
post Aug 29 2013, 04:24
Post #12





Group: Members
Posts: 1080
Joined: 11-February 12
Member No.: 97076



Also yes, MPC's decoding speed is very fast but AAC and Ogg Vorbis are almost double the speed. Am I doing something wrong in my test?

http://dropcanvas.com/vq4w7?expand

edit:
Adding more lossless in a bit.

edit2:
Done.

This post has been edited by eahm: Aug 29 2013, 04:56


--------------------
/lwAsIimz
Go to the top of the page
+Quote Post
B7k
post Aug 29 2013, 05:11
Post #13





Group: Members
Posts: 45
Joined: 11-March 07
From: cleveland,ohio
Member No.: 41371



@ eahm Maybe so on pc but I was referring to rockbox firmware for the Ipod video 5.5G it does have it's processing limitations. Interesting benchmark testing though.

This post has been edited by B7k: Aug 29 2013, 05:11
Go to the top of the page
+Quote Post
eahm
post Aug 29 2013, 05:16
Post #14





Group: Members
Posts: 1080
Joined: 11-February 12
Member No.: 97076



QUOTE (B7k @ Aug 28 2013, 21:11) *
@ eahm Maybe so on pc but I was referring to rockbox firmware for the Ipod video 5.5G it does have it's processing limitations. Interesting benchmark testing though.

Thanks. OT: To me the most interesting thing is why WAV is slower than AIFF. I need to find a book/website where I can study the decoding process, I like it more than anything else in audio.


--------------------
/lwAsIimz
Go to the top of the page
+Quote Post
rutra80
post Aug 29 2013, 11:15
Post #15





Group: Members (Donating)
Posts: 810
Joined: 12-September 03
Member No.: 8821



QUOTE (eahm @ Aug 29 2013, 06:16) *
OT: To me the most interesting thing is why WAV is slower than AIFF. I need to find a book/website where I can study the decoding process, I like it more than anything else in audio.

Little vs big endian?
Go to the top of the page
+Quote Post
DigitalDictator
post Aug 29 2013, 12:52
Post #16





Group: Members
Posts: 313
Joined: 9-August 02
From: SoFo
Member No.: 3002



IIRC, the developer, Bauschman (?), just didn't have time nor the motivation to continue with the Musepack project, and his computer broke down at the same time. People here were so frustrated (and desperate), they even offered him money for a new computer so he could continue with his work. But in the end, he abandoned the project. It was a while ago, but that's what I think happened.
Go to the top of the page
+Quote Post
shadowking
post Aug 29 2013, 13:24
Post #17





Group: Members
Posts: 1526
Joined: 31-January 04
Member No.: 11664



IMO Buschmann & Klemm where way ahead of the game and their time - We are talking circa 2000 where 128k and swooshing were commonplace. MP3 was only stable more or less at 192 CBR. VBR was hit and miss + slow, AAC was still immature, Vorbis plagued by a HF noise. MPC quality was already nearly finalized at that time. Klemm made the encoder very fast by 2002. Lame -APS was a fat lumbering elephant compared to mpc Q5.

The trouble was the format was ahead of its time. At least today with phones / tablets you can load a player that will handle mpc and other non common formats. The other problems was the two developers never had much time to finalize the format in terms of specs etc.. This is where vorbis got it right sort of but similar loss of interest happened there too till this day. Vorbis had big quality issues until aotuv. Several people tried to fix it but only aotuv developer could do it. If he didn't come along vorbis would have remained same quality till today. Anyway around 2004-5 MDT did a good job with the help of klemm and buschel - but in the gap around 2003-2005 a lot of damage was done. Without a developer / maintainer people got scared and started leaving.


--------------------
Wavpack -b450s0.7
Go to the top of the page
+Quote Post
Canar
post Aug 29 2013, 17:03
Post #18





Group: Super Moderator
Posts: 3361
Joined: 26-July 02
From: princegeorge.ca
Member No.: 2796



QUOTE (DigitalDictator @ Aug 29 2013, 04:52) *
People here were so frustrated (and desperate), they even offered him money for a new computer so he could continue with his work. But in the end, he abandoned the project.


The computer fundraising was not to coerce Klemm to continue, it was pitched more as a thank-you for his work to that point. Some were hopeful that it might catalyze him to contribute more. However, he open-sourced, and SV8 happened, and Musepack is as good as it's ever been.

Whenever I can say "Fuck compatibility, I want transparency at low bitrates", I end up going with Musepack. The fact that it's light on decoding resources is a lovely plus.


--------------------
You cannot ABX the rustling of jimmies.
No mouse? No problem.
Go to the top of the page
+Quote Post
eahm
post Aug 29 2013, 17:24
Post #19





Group: Members
Posts: 1080
Joined: 11-February 12
Member No.: 97076



QUOTE (Canar @ Aug 29 2013, 09:03) *
Whenever I can say "Fuck compatibility, I want transparency at low bitrates", I end up going with Musepack. The fact that it's light on decoding resources is a lovely plus.

"Low bitrates"? Are we still talking about ~180 kbps? I thought low are ~64/96 (Opus, AAC)?

This post has been edited by eahm: Aug 29 2013, 17:25


--------------------
/lwAsIimz
Go to the top of the page
+Quote Post
Canar
post Aug 29 2013, 17:29
Post #20





Group: Super Moderator
Posts: 3361
Joined: 26-July 02
From: princegeorge.ca
Member No.: 2796



"Transparency at low bitrates" means 140-180ish to me. Below that point there isn't transparency (except Opus and maybe HE-AAC, but the latter has never ABXed well for me). Medium bitrate would be LAME V0. High bitrate is lossless.

I'm old. So sue me. tongue.gif


--------------------
You cannot ABX the rustling of jimmies.
No mouse? No problem.
Go to the top of the page
+Quote Post
Dynamic
post Aug 29 2013, 17:59
Post #21





Group: Members
Posts: 815
Joined: 17-September 06
Member No.: 35307



I think to summarise:

Listening tests are too hard at near-transparent bitrates, so in recent years listening test bitrates have come down (128, 96, 64, 48 etc). This has led to Musepack not featuring as it's very uncompetitive below 128.

Musepack is great and very well tuned for transparent audio (Q5 or above) and still remarkably good at Q4 (128kbps or so).

Below Q4 bitrate starvation causes the quality to degrade quickly like a lot of old technology codecs like MP2, MP3 which don't have special tricks to hide degradation or to encode stereo adequately with much lower bitrate.

Ogg Vorbis has some tricks to roughly maintain the coarse critical band energy, but this was discovered late in its development so isn't implemented as efficiently as it might be, but it's pretty good down to about 80 kbps in AoTuV. Opus/CELT had this as a fundamental part of its design, implemented efficiently. Both have smart ways to encode stereo efficiently at lower bitrates.

AAC-LC has some efficiencies, making it good down to about 96kbps. Below that, the flaws in LC get rapidly worse, so HE-AAC is better by the time we reach 64kbps and we put up with Spectral Band Replication being an inexact representation of the high end. Below about 48 kbps, Parametric Stereo is also introduced to enable even lower bitrates with a good impression of bright full range sound, but certainly frequent flaws at 24 kbps.

It seems that AAC-LC and Opus are contenders for providing transparency at about 128kbps average (subject perhaps to further careful tuning).

This post has been edited by Dynamic: Aug 29 2013, 18:03
Go to the top of the page
+Quote Post
saratoga
post Aug 29 2013, 19:45
Post #22





Group: Members
Posts: 4959
Joined: 2-September 02
Member No.: 3264



QUOTE (shadowking @ Aug 29 2013, 08:24) *
IMO Buschmann & Klemm where way ahead of the game and their time - We are talking circa 2000 where 128k and swooshing were commonplace. MP3 was only stable more or less at 192 CBR. VBR was hit and miss + slow, AAC was still immature, Vorbis plagued by a HF noise. MPC quality was already nearly finalized at that time. Klemm made the encoder very fast by 2002. Lame -APS was a fat lumbering elephant compared to mpc Q5.


Buschel is still active working on Rockbox from time to time. He fixed up our MPC and AAC decoders quite a lot, which is part of why MPC is now so efficient on ARM devices.
Go to the top of the page
+Quote Post
saratoga
post Aug 29 2013, 19:55
Post #23





Group: Members
Posts: 4959
Joined: 2-September 02
Member No.: 3264



QUOTE (eahm @ Aug 29 2013, 00:16) *
QUOTE (B7k @ Aug 28 2013, 21:11) *
@ eahm Maybe so on pc but I was referring to rockbox firmware for the Ipod video 5.5G it does have it's processing limitations. Interesting benchmark testing though.

Thanks. OT: To me the most interesting thing is why WAV is slower than AIFF. I need to find a book/website where I can study the decoding process, I like it more than anything else in audio.


Where are you looking?
Go to the top of the page
+Quote Post
halb27
post Aug 29 2013, 20:40
Post #24





Group: Members
Posts: 2435
Joined: 9-October 05
From: Dormagen, Germany
Member No.: 25015



Out of curiosity I tried my worst problem samples harp40_1, herding_calls, eig, lead-voice, trumpet_myPrince with Musepack SV8 using standard quality (5).
I'm really impressed, quality is great. I did only a short ABX test and didn't succeed with any of these samples (though I do believe I can ABX eig when applying a lot more effort). For practical listening I'm totally happy with these results.

More interesting to me: my Nokia C7 smartphone was broken recently, and I replaced it with a SONY Xperia L android smartphone for which I use Poweramp for listening to music. I just realized that Poweramp plays mpc files (as do other android players). My home stereo player is a Rockbox armed Clip+, so no problems here too.

So I could use Musepack, so far. I don't care about codec development to have discontinued as the codec seems to be great. What I care about is playback support on modern devices, and this seems to look good.
Sharing music has come down for me to share music with my wife. She can't use Musepack so this is still a problem. A minor one though as she has her collection, and additions don't happen frequently.
I'm used to change loudness by modifying the mp3 scale factors according to RG value using foobar, with a higher amount of manual RG modification, and I have written a tool to take these RG values to renewed encodings. I can't migrate this process to Musepack, but as PowerAmp and Rockbox can make use of RG values for playback, I can find a corresponding solution.

I'll think about it.



--------------------
lame3100m -V1 --insane-factor 0.75
Go to the top of the page
+Quote Post
LedHed8
post Aug 30 2013, 01:48
Post #25





Group: Members
Posts: 43
Joined: 1-June 13
Member No.: 108428



QUOTE (halb27 @ Aug 29 2013, 14:40) *
Out of curiosity I tried my worst problem samples harp40_1, herding_calls, eig, lead-voice, trumpet_myPrince with Musepack SV8 using standard quality (5).
I'm really impressed, quality is great. I did only a short ABX test and didn't succeed with any of these samples (though I do believe I can ABX eig when applying a lot more effort). For practical listening I'm totally happy with these results.



Me too. I've been impressed so far using standard quality. I've resisted trying Musepack because it seemed "fringe and obsolete". It might be fringe, but there does not seem to be anything obsolete about the sound quality. From a quick couple of encodes and listens, I must encode more music and try it for a while. Even my wife, who could generally care less about my sound obsession, was genuinely impressed with the Musepack sound. biggrin.gif
Go to the top of the page
+Quote Post

3 Pages V   1 2 3 >
Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 17th September 2014 - 05:28