Help - Search - Members - Calendar
Full Version: MPC Characteristics
Hydrogenaudio Forums > Lossy Audio Compression > MPC
Ruse
A synopsis of the codec fora (this one and others) participants’ views and my own observations and testing of MusePack (mp+/c, mpegplus) versus mp3

[Encoders and settings as follows unless stated –
mp3 – LAME 3.90 Alpha 8 (6th Nov 2001, --dm-preset standard (Variable Bit Rate by default);
MusePack – Frank Klemm mppenc 0.90k, Standard (VBR by default).
Average bit rates are quoted in kbps from Helium2 v1.01)

MusePack Characteristics:

1. Clearly has an audible advantage coding music with sharp transients or impulses, which includes specifically, synthesiser-generated comb like waves, castanets, cymbals, etc. This audible advantage has a strong basis in theory, with MusePack not suffering from pre-echo. Examples of waves or tracks which demonstrate this advantage -
kalifornia-clip (by Fatboy Slim) (21 secs). [MusePack unknown setting bitrate average = 253, mp3 CBR 256. A senior contributor to tuning LAME settings has spent considerable effort tuning LAME for clips such as this, but the results are still obviously artifacted – no ABX required)
Castanets.wav (6 secs). [MusePack unknown setting bitrate average = 253, mp3 LAME VBR unknown setting bitrate average = 197]

2. Pipe instruments have clearer definition – Northumbrian pipes, Bagpipes, Ulean pipes, etc. Tracks which demonstrate this:
Soundtrack to Titanic (James Horner) - Hymn to the Sea (6:25) [MusePack bitrate average = 197, mp3 bitrate average = 183]
The Corrs - Forgiven Not Forgotten – Erin Shore (Instrumental) (0:27) [MusePack bitrate average = 222, mp3 bitrate average = 164]

It is interesting to note that MusePack forces the bitrate up on the transients and pipe sections to cope with these waveforms, whereas mp3 does not seem to do this. This is not a fault of either encoder, it is just indicative of the way the two different encoders work. The average bitrate for the two encoders at settings defined above is very similar across a number of CDs.

3. A census of those people that moderate or are respected by these fora, indicates that MusePack is regarded as of much better quality than mp3 at any bitrate above approx 160 kbps. That is, MusePack produces fewer artifacts at equivalent bitrate to mp3, or even compared to mp3 at 320 kbps CBR or ‘insane’ VBR settings. MusePack demonstrates transparent quality at Xtreme setting for 99.9% of samples.

4. To achieve a similar quality, MusePack uses fewer bits (disk space) than mp3.

5. Excellent psycho acoustic tuning. There is little need for tweaking of command lines such as is the case with LAME.

6. MusePack encodes at approximately 3 to 4 times the CPU speed of LAME using the quoted settings. MPC is the fastest encoder for the level of quality that it provides.

7. MusePack has a low decoder complexity compared to mp3 (and AAC).

8. There is no maximum bitrate imposed on MusePack, compared to the 320 kbps of mp3. MusePack can be used at bitrates from approx. 160 up to approx. 700 kbps. Muse Pack can be used as a very high bitrate archiving format if required.

9. Software support is limited, but suitable programs such as WinAmp and Helium2 are available.

10. MusePack is free.
spase
just want to add that mpc also performs well at about 128 kbps as could be seen in ff123's listening test (forgot url... maybe someone can provide it...)
kennedyb4
There has been a ton of excellent reasons to use MPC. I would have been using it for ages except for a couple things.

MP3, while not perfect maybe, still sounds damn good on most material with these HQ presets. And the tweaking is fun too. MPC presets have less "play" attraction. (flame me, please)

Still, personally I would switch immediately if I could get a portable MPC player.I am still waiting to actually see these firmware upgrades for stuff like the Rio Volt or Pocket Concert.
I have heard MPC transcodes very well but no listening tests on this.

So the hold-up still seems to be compatability. I decode to wav with HRTF's and a crossfeed filter currently so it doesn't matter to me now, but it isn't hard to picture a nice quality portable with 256 of ram on a little belt clip either in the near future.
jkml
I read somewhere that one day the MPC encoder will not be freeware anymore, although the decoder will remain free. Is it true?
spase
probably.... its all up to andree and his crew tho...

he DID have to buy patents from phillips or someone to use sub band coding...

id be happy to pay him some money for his work tho...

it blows mp3 away!
JohnMK
QUOTE
Originally posted by kennedyb4
I have heard MPC transcodes very well but no listening tests on this.


So the theory goes. In the absence of listening tests that's all we have to go by. I am concerned that masking becomes more complex, but that's just inherent. Perhaps if we could tell an encoder (mp3) that it's receiving previously-compressed material, it might produce a more optimal transcode. So far, though, I haven't been able to distinguish between direct CD -> mp3 versus CD -> mpc -> mp3, but I know there are far more discriminating ears than my own.

Ringa ding Xing.

John
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.
Invision Power Board © 2001-2008 Invision Power Services, Inc.