Topic: Why is ripping so slow? (Read 5013 times)

fex · « **on:** March 15, 2003, 04:19:20 am »

Why is ripping so slow within MC? With MusicMatch I rip over 40x, in MC I only get 13x. Is there any explanation?

Fex

KingSparta · « **Reply #1 on:** March 15, 2003, 04:47:51 am »

yes

1. depends on your settings
2. MusicMatch cuts out more info so the sound is not as good
3. some users use the command line and use the lame encoder. they say if you use the right command line it is faster.

fex · « **Reply #2 on:** March 15, 2003, 06:57:41 am »

Thanks. Hear absolutely no difference with my old ears at 128 kbps.

But I will try to find out the right lame settings (speed against quality) for my girlfriend. Don't want to loose her...

cct1 · « **Reply #3 on:** March 15, 2003, 07:33:25 am »

You might want to try this (thanks to Doof for this):

Go to the "encoder" options, and for the encoder, select the the "MP3 Encoder VBR".

Underneath that, in quality, pick "Custom" and click on advanced. You should see "--alt-preset fast standard" in the command line--click ok and try ripping.

This uses the LAME encoder and is ripping lightning quick for me, with excellent quality. May be worth a try for you.

michel · « **Reply #4 on:** March 15, 2003, 08:29:29 am »

Slow rip is not only matter of conversion speed. It is also matter of extraction (from CD) spped.

According to your settings, cd extraction can be more or less fast.

MC 'digital secure' or EAC 'paranoia mode' (I don't remember the exact name) can be very slow. On the other side they are more accurate.

Check which mode is selected into MC.

NoCodeUK · « **Reply #5 on:** March 15, 2003, 08:41:40 am »

Nice One! This upped my ripping speed by about 500% and the file quality is as good or as better as the settings I was using (VBR High, High Quality)

Adam

cct1 · « **Reply #6 on:** March 15, 2003, 09:08:27 am »

Quote

Slow rip is not only matter of conversion speed. It is also matter of extraction (from CD) spped.

According to your settings, cd extraction can be more or less fast.

MC 'digital secure' or EAC 'paranoia mode' (I don't remember the exact name) can be very slow. On the other side they are more accurate.

Check which mode is selected into MC.

True, but in my experience, the compression has always taken longer than the ripping, even in digital secure mode. I use the MC9 ripper in digital secure, and the LAME encoder with those --alt-preset fast standard settings--and it is remarkably fast--give it a shot and see what you think!

rocketsauce · « **Reply #7 on:** March 15, 2003, 10:41:01 am »

Quote

2. MusicMatch cuts out more info so the sound is not as good

?

GrumpyNick · « **Reply #8 on:** March 15, 2003, 10:57:18 am »

In addition, check whether "Analyze audio during ripping" is switched on under Tools...Options...Device Settings. This slows down ripping a lot - but saves having to analyze the files afterward.

Nick

fex · « **Reply #9 on:** March 15, 2003, 12:08:45 pm »

Quote

yes

... MusicMatch cuts out more info so the sound is not as good

Did a simple test. Ripped "Don't Bring Me Down" by Electric Light Orchestra (duration on CD 4.03) with MM and MC.
Result:

3.724 MB with MusicMatch;
3.722 MB with MediaCenter;

both with 128 kbps and constant bitrate.

What is MusicMatch "cutting away"? Why MusicMatch is 3x (!) faster ripping than MediaCenter? Hate to use a second program besides MC to rip (as I do now), so any further (and logical) explanations welcome.

Fex

rocketsauce · « **Reply #10 on:** March 15, 2003, 12:24:45 pm »

It's kind of a two step process.

As mentioned in one of the posts above, if you have MC set to rip in Secure Mode it will almost certainly be slower than MMJB at ripping the audio from the CD but more accurate.

Another factor is that MMJB and MC use different MP3 encoders. MC uses the LAME encoder. I'm not sure which particular one MMJB uses. However, MMJBs encoder is simply faster at encoding than LAME, but it is generally accepted that LAME creates files of a higher quality, especially at higher bitrates. Since you are encoding at 128kbps, though, there is probably not much noticable difference in the sound quality between the two.

Rob

nekura · « **Reply #11 on:** March 15, 2003, 12:31:46 pm »

Quote

I'm not sure which particular one MMJB uses.

I'll take Fraunhofer for $500. =D

fex · « **Reply #12 on:** March 15, 2003, 12:49:40 pm »

Quote

...but it is generally accepted that LAME creates files of a higher quality

MMJB uses version 3.2 of the Fraunhofer encoder (by the way, they did MP3). And I really hear no difference to LAME with 128 kbps. But I feel the missing speed from LAME (Lama?).

Could it have something to do that Fraunhofer is not for free?

I love to save time, and with the given "explanations" in this thread so far I don't come around to use MMJB for ripping in the future. And I hate this besides MC.

Fex

fex · « **Reply #13 on:** March 15, 2003, 12:59:48 pm »

Quote

I'll take Fraunhofer for $500. =D

By the way: You would get "MMJB Plus" for USD 9.99. With the Fraunhofer encoder included. What did you pay for MC (with free LAME?). Don't forget, I payed also for MC, and I exactly know why! (not for LAME...

).

Fex

cct1 · « **Reply #14 on:** March 15, 2003, 01:24:13 pm »

Fex: try the --alt-preset fast standard setting in the MC9 LAME encoder and see what you think. The file size will be bigger, the kbs will average out around 190--but to me there is a noticeable difference in sound. If you absolutley do not want the larger file size, try turning off the analyze audio--but this is a nice feature to have turned on, so you can use it to make smartlists.

fex · « **Reply #15 on:** March 15, 2003, 01:43:41 pm »

I'll try it and let you know about my experience. Will test the results on my home stereo. Tanks for the suggestions so far.

Fex

fex · « **Reply #16 on:** March 15, 2003, 03:01:56 pm »

Quote

Fex: try the --alt-preset fast standard setting in the MC9 LAME encoder and see what you think...

Ok, dit it. Ripped the album "Naked" from "Blue Pearl", 1990, once with MC9 LAME (alt preset fast standard) and once with MMJB (192 kbps, vbr 60%)

- with MC I reached top speed 5.2x, whole album in 11 Min. 4 Sec.
- with MMJB I reached speed 30.4x, whole album in 2 Min. 40 Sec.

Both result in files around 190 kbps. Size is more or less equal. Sound is better than 128 kbps (logical), but for me no difference between MMJB and MC.

How can the amazing speed difference be explained (that was the question...)?.

Fex

cct1 · « **Reply #17 on:** March 15, 2003, 03:43:08 pm »

Hmm..Now you've got me, unless it's due to the "analyze audio" (this gives you things like "intensity" and BPM {Beats Per Minute}--you can make some pretty cool smartlists with these.

If you rip/encode with mmjb, you can still "analyze audio" through tools, but it takes awhile (one of the reasons I use MC9 to rip/encode instead of EAC/LAME, which is what I used to use).

KingSparta · « **Reply #18 on:** March 15, 2003, 03:55:35 pm »

>> How can the amazing speed difference be
>> explained (that was the question...)?.
It's Just Faster

However there was some research done that checked the freq range of the ripped songs and they stated Lame was better.

Me and you can't hear it but others can.

I think Zrocker Knew what the site link was and you can see the drop off of freq. compared to the Lame encoder.

It is not going to get faster no matter how many times you ask.

If you find an encoder that works better as an external encoder let us know.

GoGo Encoder also worked and is a multi thread Encoder

Some Reading:

http://users.belgacom.net/gc247244/index.html (With Bar Graphs)

If you look on the graph Fhg is the worst at 320bps than any of the Lame, Lame at 320bps is close to the orginal.

Quote

Findings:

The best Fraunhofer products are less accurate than the Lame products.
The JS modes, up to 256 of course, slightly outperform their respective S modes.
The vbr signal is also up there.
Regarding this Lame 3.80 graph, and not the totally different 3.70 graph, my intuition would say that at 256kbit/s you'd better go for Stereo to avoid the slight possibility that JS might introduce artifacts.
For VBR I advise you not to use S instead of JS, because this will add 30 ~> 40kbit/s to your average bitrate, defeating the whole purpose of having a decent quality/size ratio. Now, with LAME 3.80, even the most sceptic of my correspondents had to admit they could not find any more artifacts using my suggested "-V1 -b128 -mj -h" setting.

--------------------------------------------------------------------------------
22 feb 2001: In case all these frequency analysis confuses you, leaves a careless or superficial impression or you want to know in more detail how these were conducted: feel free to check out the "analysis walktrough".
--------------------------------------------------------------------------------
Conclusions:

I Radium FhG Codec in HQ & Stereo,
Newer FhG encoder found in MM and Nero in "lowest quality" & Stereo,
Blade 0.9.3.6 in 256kbit/s, Stereo and
Lame vbr JS or Lame 256 cbr JS
reproduce enough music quality to be used for archiving music in MP3 format.
II LAME >= 3.85 -b256 -ms -h 256kbit/s Stereo is distinctly the most accurate mp3 encoder (Best archiving for cbr)
III LAME = 3.85 -V1 -b128 -mj -h -q1 VBR is by far the best overall archival mp3 encoder if size does matter (highest quality/size ratio)

http://www.modatic.net/audio/mp3_encoder_comparison.php

http://www.cdfreaks.com/document.php3?Doc=45

rocketsauce · « **Reply #19 on:** March 15, 2003, 05:35:21 pm »

Quote

I'll take Fraunhofer for $500.

That's what I thought,

but I couldn't check since I haven't been able to get MMJB to run on my system for about 9 months now.

Quote

- with MC I reached top speed 5.2x, whole album in 11 Min. 4 Sec.
- with MMJB I reached speed 30.4x, whole album in 2 Min. 40 Sec.

That's a good speed for LAME. On my system it encodes at about 1.8x.

However, you also need to take into account the ripping method of both programs. MMJB probably defaults to using a "burst mode". This will definitley be faster than MCs secure mode.

To test differences in the ripping speed (as opposed to ripping/encoding speed), try copying a disc with each program set to rip to uncompressed WAV instead of MP3. This will help give you an idea of how much of the speed differences are due to each step of the process.

Quote

How can the amazing speed difference be explained (that was the question...)?

There are several different MP3 encoders. LAME and Fraunhofer are only two. They all have their good points and bad points. I don't know the technical reasons why Fraunhofer is faster, but it is. If you're really interested in finding out, I bet you could post over at Hydrogen Audio and someone would be able to offer more indepth details.

http://www.hydrogenaudio.org

Of course, ultimatly, if you can't hear any quality differences between the two encoders you should use the faster one.

Rob

JohnT · « **Reply #20 on:** March 17, 2003, 03:58:47 am »

Quote

Ok, dit it. Ripped the album "Naked" from "Blue Pearl", 1990, once with MC9 LAME (alt preset fast standard) and once with MMJB (192 kbps, vbr 60%)

- with MC I reached top speed 5.2x, whole album in 11 Min. 4 Sec.
- with MMJB I reached speed 30.4x, whole album in 2 Min. 40 Sec.

Both result in files around 190 kbps. Size is more or less equal. Sound is better than 128 kbps (logical), but for me no difference between MMJB and MC.

How can the amazing speed difference be explained (that was the question...)?.

Fex

The Fraunhofer encoder simply has faster algorithms. I think they have concentrated on the speed aspect more than the quality aspect, although if you can't hear the difference it's a moot point. You could do me a favor if you would conduct a test on your system using WMA as the encoder to see if there's some other speed problem going on that we could fix in Media Center. Just set up both MMJB and MC to rip at highest speed (digital large buffer in MC) and turn off audio analysis in MC. Choose wma at the same encoding rate in both programs, then tell us the timings for a rip.

Thanks,
John T.
JRiver, Inc.

fex · « **Reply #21 on:** March 17, 2003, 11:03:56 am »

Quote

...You could do me a favor if you would conduct a test on your system using WMA as the encoder to see if there's some other speed problem going on that we could fix in Media Center. Just set up both MMJB and MC to rip at highest speed (digital large buffer in MC) and turn off audio analysis in MC. Choose wma at the same encoding rate in both programs, then tell us the timings for a rip.

Ok, here the results concerning WMA:
Ripped the album "Naked" from "Blue Pearl" again.

64 kbps:
- MMJB: Top speed 21.4x in 3 Min. 33 Sec. Result regarding first file: 1.99 MB with 65 kbps (processing level at "normal").
- MC: Top speed 26.1x in 2 Min. 37 Sec. Result regarding first file: 1.97 MB with 64 kpbs.

160 kbps:
- MMJB: Top speed 19.0 in 3 Min. 45 Sec. Result regarding first file: 9.89 MB with 160 kbps (processing level at "normal").
- MC: Top speed 23.8 in 2 Min. 37 Sec. Result regarding first file: 9.87 MB with 160 kbps.

So MC is quite faster than MMJB ripping WMA. So you did your job, there is not really a speed problem in this point!

Fex

Quote

Media Center Registered 9.0.129 -- C:\Programme\Media Center\

Microsoft Windows XP Workstation 5.1 Service Pack 1 (Build 2600)
Intel Pentium 4 1999 MHz MMX / Memory: Total - 523 MB, Free - 209 MB

Internet Explorer: 6.0.2800.1106 / ComCtl32.dll: 5.82 (xpsp1.020828-1920) / Shlwapi.dll: 6.00.2800.1106 (xpsp1.020828-1920)
Shell32.dll: 6.00.2800.1145 (xpsp2.021108-1929) / wnaspi32.dll: Internal ASPI Layer

Ripping / Drive F: Copy mode:ModeBurstBigBuffer CD Type:Auto Read speed:Max
Drive G: Copy mode:ModeBurstBigBuffer CD Type:IDE Read speed:Max
Digital playback: Yes / Use YADB: Yes / Get cover art: Yes / Calc replay gain: No / Copy volume: 32767
Eject after ripping: Yes / Play sound after ripping: Yes Soundfile: chord.wav

Burning / Drive G: YAMAHA CRW-F1E Addr: 1:0:0 Speed:10 MaxSpeed:44 Use MJ Engine:Yes
Test mode: No / Eject after writing: Yes / Direct decoding: Yes / Write CD-Text: No
Use playback settings: No / Normalization: None

fex · « **Reply #22 on:** March 17, 2003, 11:08:49 am »

P.S.: Hate my 20GB-iPod and my car-stereo. Both don't understand WMA-format...

JohnT · « **Reply #23 on:** March 17, 2003, 12:16:00 pm »

Thanks for running the test. We'll keep looking for a faster mp3 encoder that we could offer as an alternative.

Markeau · « **Reply #24 on:** March 17, 2003, 12:39:45 pm »

After reading posts that MMJB rips have a steep cutoff at 16KHz I looked at some MMJB rips with Cool Edit's freq analyzer. Indeed all MMJB rips (320 CBR, Very High and Max Bandwidth slider pushed all the way to 22KHz) had a very steep cutoff right at 16KHz. Using MC's native LAME (320 CBR, High) the rips go all the way up to about 21KHz.

fex · « **Reply #25 on:** March 17, 2003, 01:47:00 pm »

Quote

After reading posts that MMJB rips have a steep cutoff at 16KHz I looked at some MMJB rips with Cool Edit's freq analyzer. Indeed all MMJB rips (320 CBR, Very High and Max Bandwidth slider pushed all the way to 22KHz) had a very steep cutoff right at 16KHz. Using MC's native LAME (320 CBR, High) the rips go all the way up to about 21KHz.

Yup. All methods in Cool Edit's analyzer (Blackman/Harris, Welch or the ohters) show this. Could be (or is for sure) one of the reasons for the speed difference. When 16Khz is to low for cut off, maybe 18Khz could do the job? Nothing to decide for my ears...

KingSparta · « **Reply #26 on:** March 17, 2003, 01:54:59 pm »

Since a Human Ear can Hear from 20-20,000Hz any time you cut some out of that range it will effect the sound quality.

Human speech is from -<1,000 - ->4000hz

sertua · « **Reply #27 on:** March 18, 2003, 04:33:15 am »

On codecs

Looking at graphics is not the most efficient mean to evaluate codecs. Info from king is interesting but most points out to non-updated and controversial r3mix-related analysis (http://www.r3mix.net/). For another point of view:
http://www.hydrogenaudio.org/index.php?act=ST&f=1&t=4298&hl= (originated by one of chicoself's first post on hydrogenaudio

) and http://www.hydrogenaudio.org/index.php?act=ST&f=1&t=4015&hl=r3mix&s=2d24894f78db0618326dea22d5f4e0a1 (check Flloyd's answer particularly and later Dibrom's)

It is normally accepted to say that LAME and Fraunhofer's FastEnc (used since MMJB 6.1) have similar quality at 128 kbits/s; some background:
http://www.ff123.net/cbr128.html and http://www.ff123.net/index.html

Anything at a higher bit rate with Fraunhofer's FastEnc is inefficient. FH was optimised for speed (and its development stopped years ago) while LAME is optimised for quality (and still in development, making use of the latest psychoacoustic models). If you want better sound quality with mp3, a slower encoding process shouldn't be too much of a high price to pay. Other types of lossy codecs will provide for more quality and speed (mpc, ogg, aac, but definitely not wmv), but at the price of hardware compatibility (at least now).

Also, according to the LAME development team, the next release of LAME (3.94, - curently in late alpha stage-) is going to have noticeable speed increases.

If you want to learn more about audio-related stuff, a good and up to date place to start is here:
http://www.hydrogenaudio.org/index.php?act=ST&f=1&t=4917&

and after reading the FAQ the fine people at hydrogenaudio will be happy to help you out.

On MJ/MC

A way of easily improving speed in MJ/MC would be to allow for parallelized ripping/encoding, as previously discussed and many times requested:
http://www.musicex.com/cgi-bin/yabb/YaBB.cgi?board=beta;action=display;num=1038804011
and http://peakin.com/music/rip_comparison.html (RemyJ's test?)

This is especially true for older CPU (like my PIII-700), as ripping is much faster in my case (6x in secure mode) than the encoding process (1,5x on average using alt-preset standard). At the fastest setting (rip and encode) my drive makes very long pauses waiting for the CPU to encode after a song is ripped, and in turn my CPU will then wait for the drive to rip before encoding. If MC/MJ could have the encoding and the ripping work simultaneously, the overall speed would be faster by just eliminating delays. The lack of such feature is the only thing that prevents me from abandoning Exact Audio Copy for most of my ripping needs.

INTERACT FORUM

Author Topic: Why is ripping so slow? (Read 5013 times)

fex

fex

fex

fex

fex

fex

fex

fex

fex

fex