Did some timings for different encoders, just to see what the difference was. Used Lame at --alt-preset standard -Z for a lossy file size reference and FLAC setting 5 for a lossless file size reference, I tried to use settings on the other encoders to get a comparable file size (or near). This was not really about quality although most of the settings are recommended ones from this forum, rather the difference in timings between the encoders for comparable file sizes on differing computers especially from the processor point of view. All files were encoded using Speek's frontends, bar Monkey's (I just used the generic GUI) If this has been done before I apologise but I did search and could not find anything quite like what I was after.

Computer 1- Windows XP Home Edition, 512 RAM, 1400mhz Athlon (not XP).
Computer 2- Windows 98se, 192 RAM, Pentium II 300mhz.
(Sorry, I don't have a computer running Linux or other OS)
Both computers had a fresh boot prior to encoding so there was very little processor usage going on in the background, I don't have heaps of programs running at windows start up so any usage would just be Windows default.


Song used, Doves Caught by the River 5 mins 55 secs (thought it was a good average song at this running time, music is not too heavy nor too soft)

Windows XP:
Ripped WAV using EAC (secure mode) No errors reported- 1 mins 23 secs @ 59.7mb

Psytel AAC v2.15: --normal / 2 mins 43 secs / 7.39mb
--extreme / 2 mins 43 secs / 9.24mb
LAME 3.90.2 (HA Recom.): --alt-preset standard -Z / 1 min 09 secs / 10mb
MPC Encoder v1.14: -quality 6 --xlevel (xtreme) / 26 secs / 8.66mb
-quality 7 --xlevel (insane) / 26 secs / 9.81mb
Ogg Vorbis GT3b1: -q6 / 40 secs / 8.71mb
Ogg Vorbis oggenc: -q7 / 42 secs / 9.05mb
FLAC 1.1: setting 5 / 11 secs / 42.4mb
Monkeys 3.97: Fast / 7.78 secs / 41.7mb
Wavpack: Lossless Normal / 7.73 secs / 42mb

Windows 98se:
Ripped WAV using EAC (secure mode) No errors reported- 1 min 39 secs @ 59.7mb

Psytel AAC v2.15: --normal / 11 mins 09 secs / 7.39mb
--extreme / 11 mins 01 secs / 9.24mb
LAME 3.90.2 (HA Recom.): --alt-preset standard -Z / 5 mins 32 secs / 10mb
MPC Encoder v1.14: - quality 6 --xlevel (xtreme) / 2 mins 50 secs / 8.66mb
-quality 7 --xlevel (insane)/ 2 mins 51 secs / 9.81mb
Ogg Vorbis GT3b1: -q6 / 4 mins 07 secs / 8.71mb
Ogg Vorbis oggenc: -q7 / 5 mins 55 secs / 9.05mb
FLAC 1.1: setting 5 / 1 min 10 secs / 42.4mb
Monkeys 3.97: Fast / 40 secs / 41.7mb
Wavpack: Lossless Normal / 37 secs / 42mb

Thats it, My main reason for doing this was to see what would be the fastest format to use on my old PII bearing in mind quality and file size. So it looks as though MPC is the big winner for lossy and Wavpack for lossless. Hopefully this will maybe be of some help to anyone else trying to decide which format to go for.

Cheers

Edit: Updated FLAC figures to v1.1.

Edit: Updated Monkey's and Wavpack settings to give a more fair comparison (in relation to filesize) to FLAC.

Edit:

For all who are interested, same song only ripped/ encoded straight from CD, as is MP9's want. I could not get a setting for a file size around the same as the others for lossy so I encoded it using the two highest settings. Encoded using WMA 9 format (straight not Pro) with Media Player 9, copy protection disabled.

Windows XP:
WMA Var 240-355kbps / 26 secs / 12.9mb
131-215kbps / 27 secs / 7.5mb
WMA Lossless / 30 secs / 41.4mb

Windows 98se:
WMA Var 240-355kbps / 3 mins / 12.9mb
131-215kbps / 2 mins / 7.5mb
WMA Lossless / Not available on Windows 98 via MP9

I was not aware that Lossless was not supported on 98 by MP9 until I tried to select it in MP9's settings, not a good performance on 98se anyway. dBpowerAMP Music Converter, with the WMA plug-in enables the use of WMA lossless on Windows 98. I tried it with a WAV but was not sure if the settings were the same as MP9, anyway it encoded in 1 min 28 secs with a file size of 42.1mb.

Thanks again.

I use MPC and LAME APS. My computer is and AMD Atlhon XP 1800+ (1533 Mhz) 320 Mb RAM, 30 GB 7200 HD, Win XP Pro. The speed test was:
Wav file lenght 6:00
MPC: 22.5 secs (around 16x) (--quality 6 --xlevel)
MP3: 55 secs (around 6.5x) (--alt-preset standard -Z)
The files were encoded with EAC

Am I missing something, or is FLAC up to 1.10 now?

Yeah, your right . The FLAC codec used was from the FLAC bundle I downloaded at RareWares, thats what I have been using up until now. Thanks for the reminder

Just one clarification: It is normally the default mode of WavPack (not "high") and the "fast" mode of Monkey's Audio that produce file sizes closest to FLAC. And if speed alone is the criteria, WavPack's "fast" and "very fast" modes beat all of them:

http://web.inter.nl.net/users/hvdh/lossless/All.htm

I was wondering that too. Why is FLAC using it's normal setting, while others use the high setting?

You should either use FLAC at 8, or WavPack and Monkey's at normal, to get fair results.

Interestingly, according to that comparision, WavPack normal and Monkey's fast are both faster and compress better than FLAC at 5.

No, because you used one of the slowest AAC codecs available.

If you want a fast one, try either PsyTEL FastEnc (especially at these high bitrate settings), Nero AAC, QuickTime 6.1 Pro or the new FAAC v1.16 at -q 125.

So if someone also considers other criteria than plain speed (e.g. hardware compatibility or current and future industry standards like MPEG-4), he should not go for MPC when choosing a lossy archiving format.

And last but not least you shouldn't exclude WMA9 from such a comparison, because it can sound very good at these bitrates and also at lower ones.

Faster at encoding, yes, but not decoding (WavPack is pretty close though). It's a little unclear from the original message if both are important or not. Even though mekon21 only mentions encoding times, I would guess that "the fastest format to use on my old PII" would consider both.

Josh

The thread is titled 'encoding speeds'.

I chose FLAC at setting 5 because above this I can see no difference in the file sizes, it only increases the compression time (in my view this is it's high setting). Thats why I chose high on Wavpack and Monkeys, there is not a massive difference in file sizes but I agree these are not the best for encoding speed.

Regarding Monkeys
XP: Fast / 7.78 secs / 41.68mb
98: Fast / 40 secs / 41.68mb

Wavpack
XP: Normal / 7.73 secs / 42mb
98: Normal / 37 secs / 42mb



I never said this was an exhaustive test, please feel free to add your own speed test info. So that others could make a better choice.



If it was hardware compatability alone then the choice would be simple, MP3 or WMA. Can you say with 100% certainty what future industry standards will be ?
MS is pushing WMA, Sony is pushing ATRAC. I for one don't know, all this post was meant to be was a bit of info, surely anyone looking for a compression method would have enough sense to do a bit more research than take my word for it.



As for WMA I do not discount it, I have used it in the past for my flash player. It just doesn't seem to get much attention on these pages. I think it was rjamorim who suggested in another thread that there could be an "Others" section, I think that is a good idea, IMHO.

Cheers

Speed alone wasn't the criteria, Wavpaks "fast" and "very fast" modes are what they say they are but the file sizes were both over 44mb so I did not use them.

Cheers again.

Thanks for reporting your findings, I at least found the speed comparisons between 98 and XP interesting.

Thanks. I'm trying hard to push this idea to the authorities.



Right, but WavPack "normal" is still faster AND smaller than FLAC. That's why I don't understand why didn't you choose it:



Regards;

Roberto.

Like quality, currently available gapless encoding, and currently available and well supported tagging?

There are more factors to consider then just "industry standards."

That's true.

BTW, something I caught myself wondering: is WMA gapless?

I think this was an error on my part, as I had taken FLAC at it's "best" and also taken Monkey's at high (which I also took some "flack" for, excuse the pun) I automatically took Wavpack at high as well. It was not my intention to give false results, although I have to admit it was interesting to see Wavpack and Monkey's give back smaller file sizes at higher settings with only a little time penalty compared to FLAC at 5.

Cheers

Time to update the first post, maybe?

interesting, though ultimately pointless for me since I don't use WMA, question...I just ripped The Crystal Method - Community Service, a DJ Mix album, with WMP9 and WMA to see just this. To my ear, it sounds like it's gapless. All the tracks seem to blend together just fine. Wow, WMA has a good feature!

Very interesting. Thanks for testing it, _Shorty.

Done deal B)



This is one reason why WMA (and others) could have a more centralized section, like it or not it is one of the major mainstream encoders that a lot of people have on their computers by default. It is a lot of peoples introduction to compression. Discussing it enables us to get the best out of it, however short of perfect that is.

Sorry if this is a bit off topic

My computer is the same as your No.2 (except for 128 SDRAM, 7200rpm HD and SCSI) and I can confirm your times for mp3 and mpc. It's running somewhat slow right now because my 60 GB Data Drive has been nonstop for 3 solid weeks crunching all this out. But I doubt there would be any great difference if I were willing to take the time to optimize.

I have just encoded 2000 files 3.90.2 --alt-preset standard -Z and 1.14 --quality 8 --xlevel. As your results indicate, I've noticed virtually no difference in mpc times between --quality 6 and --quality 8.

One thing I should point out is that having also encoded the same files with 3.92 --alt-preset extreme -Z, 3.90.2 --alt-preset standard -Z was considerably slower on my machine - at least a third. But I'm sure some of that was due to fragmentation of the data drive since extreme was 1st and standard was last (about 6th) - I just don't know how much..

But your times look good from my experience.

I agree with you Garf, It's like an thread about lossless encoding

Yes!!!

Have you tested it, like _Shorty?

Please, post your test methodology and results.