Topic: rip wav file compared to eac and itune (Read 10955 times)

chew.isaac · « **on:** July 06, 2011, 06:35:39 pm »

Hi,

I rip track 8 of a "This is K2 HD Sound" CD, "Heart of Glass" by Nah Youn Sun. I rip the same song using J river, itunes and eac separately.

Then I open up the 3 wav files of the same song using Audacity and I zoom in once. Then I use windows "alt-tab" function to toggle between any two wav files. By doing that, I found that by looking at the wave forms, j river and itunes produce identical waveforms. But they are diff from eac's. The difference is not huge.

My driver has "accurate stream" but no cache. I wonder what causes this difference. It is a good disc and encounters no errors detected in EAC when ripped.

I am not sure if I will be able to hear the difference and I don't know which is the more accurate one. EAC or J river/itunes. I will do a ABX tonight using foobar when I get home as I probably can pick up better using my main system at home. I have been quite a victim of placebo.

Cheers.

Alex B · « **Reply #1 on:** July 06, 2011, 06:51:30 pm »

Probably you are seeing an offset difference (assuming you have EAC set to "correct" the offset). An offset difference does not affect audio quality.

chew.isaac · « **Reply #2 on:** July 06, 2011, 08:29:24 pm »

By offset is it like the one wave starts playing the music slightly later than another wave? All 3 wav files that I have are in sync (pardon my non technical vocab). I see information difference. I did my Masters in Engineering on Adaptive filtering and I custom programmed a FPGA to do adaptive noise cancellation. So I do see wavforms on a daily basis at that time so I am mindful whether we are seeing additional information. But I am not a sound engineer and I am not skilled in handling wav files professionally. I believe for someone who is working on sound professionally they would have much better software tools. Has anyone look at the wav files of a song rip using different software. I did see a site where a user did and he says that all his wav files ripped by different softwares (I believe he covered the 3 here that I am comparing) are all the same.

MrC · « **Reply #3 on:** July 06, 2011, 09:09:14 pm »

Since 2 of 3 match, try looking at the settings for the non-matching software, EAC. It has "Offset Technology":

See: http://www.exactaudiocopy.de/en/index.php/overview/basic-technology/offset-technology/

Matt · « **Reply #4 on:** July 06, 2011, 09:37:51 pm »

It's worth mentioning that there's no "correct" offset.

Offsets cause the track boundaries to move a little (which explains track comparison differences), but it doesn't change the gapless nature of the transition between tracks or the actual audio data.

MrC · « **Reply #5 on:** July 06, 2011, 09:50:15 pm »

I think my use of double quotes was a bit too subtle to imply a little ribbing of the touted "Offset Technology".

chew.isaac · « **Reply #6 on:** July 07, 2011, 01:21:52 am »

. I see differences in the data. Not just an offset. I just went to verify again. For offset, when toggling from one to another we should see the wave progressed in its entire waveform. But when I toggle the two wav graphs (one ripped by EAC, one ripped by j river/itunes), most of the data points remain the same, and the graphs does not shift to the left or right at all when toggling between the two. What is observed is that at different parts of the wave form, the amt of information changes. I have no idea how to "take out" the waveform in raw format from the cd and view it so I have no reference for now which is more faithful to the original. Will we be able to tell by comparing how it sounds? I don't know but I will try when I go back. Listening is subjective at best and placebo got me many times. ABX for me is a must. I just can't find any objective article on this. If I am a sound engineer, verifying this must have been such a piece of cake.

chew.isaac · « **Reply #7 on:** July 07, 2011, 01:24:09 am »

Reading matt's email again, there is a possibility I still don't understand what an offset means. I am not sure. Let me go find out more. If an offset is able to create data differences in a data file, that could be the reason.

chew.isaac · « **Reply #8 on:** July 07, 2011, 01:34:44 am »

Hi guys, I went to read up a bit on offset. My understanding of what offset is is correct. So what say earlier is really what I know for now.

Vincent Kars · « **Reply #9 on:** July 07, 2011, 02:56:37 am »

If you understand what offset is, then you will also understand that for a good comparison the tracks must be time aligned.
The trick is:
Load both rips in an audio editor
Time align
Subtract the tracks.
If the result differs from zero, the tracks are different.

glynor · « **Reply #10 on:** July 07, 2011, 10:27:51 am »

Just to double check the "silly factor"... Do you have Secure Ripping enabled in MC? It is NOT turned on by default.

If secure ripping mode is disabled, the ripping will act much more like what iTunes does by default. With secure mode turned on, MC will act basically like EAC (without the offset stuff).

chew.isaac · « **Reply #11 on:** July 07, 2011, 06:35:32 pm »

Thanks for the suggestions.

Vincent: I tried to invert one waveform and then render and mix both wave forms to achieve subtract in audacity but I failed to cancel them out. Though the jriver and itunes wave forms look exactly the same to my eyes by looking at them (toggling views). I could be doing something wrong.

Glynor: Yes, I enabled secure rip on all 3 softwares. When I just do a simple toggle on win7, I can see all 3 rips of the same song are time aligned. J river's is the same as itunes. And both are different from eac's. Eac's ripped wav is time aligned to jriver/itunes' as most points are exactly the same, I just see data difference when I toggle the views, and there is not wav form shifts on a whole.

Anyway I went back to listen. My speakers are Adam Compact Classics (powered speakers) and I use audiolab cdq. I can't tell a difference between the rip by EAC and the rips by JRiver/itunes. I know the wavforms are different but I can't hear them.

I tried to have bit accurate output from itunes by adjusting the audio properties in quicktime. I did achieve a 16 bit/44.1khz signal in my dac. But on ABX I can clearly hear the difference between itunes playback and J river (on wasapi event - which I believe is Vincent's preference as well). So I am definitely sticking with J river.

chew.isaac · « **Reply #12 on:** July 07, 2011, 07:46:00 pm »

How different does itunes playback (set to "bit accurate" via quicktime audio properties) in win7 vs J river (wasapi event)? on the song "Alfie" (track 6 of Audiophile Voices III), the singer sounds more nasal when playback in itunes. Using J river, the voice is more natural and sounds nicer to me. That's important. What's the use of bit accuracy if it does not sound nice to our ears? Then I compared it to cd playback. Indeed. JRiver playback is closer to cd playback. Itunes playback is still so different from real bit accurate playback that it becomes audible to me. By the way I did ABX tests to come to the above conclusion, as I suffer from placebo (being an audiophile), as I tend to imagine differences rather than really hear them.

Vincent Kars · « **Reply #13 on:** July 08, 2011, 03:12:09 am »

I' not familiar with iTunes so the usual grain of salt applies.
iTunes uses QuickTime.
QuickTime can use WASAPI but as far as I know only in shared mode.
As a consequence all audio goes through the Win mixer.
The Win mixer dithers the signal.
As you are playing 16 bits, this might be audible

Beside the QT settings, do check the Win audio settings.
They should be equal in bit depth and sample rate.
If you play Redbook do check that both are set to 16/44.1

You might try to compare:
iTunes and MC using DS (Direct sound)
MC using DS and using WASAPI (exclusive mode)
It is possible that the differences are simply due to the driver used (DS or WASAPI).

But I must admit, there are a lot of people who do think MC simply sounds best.
An example: http://www.soundtrackforum.net/viewtopic.php?f=4&t=325

Frobozz · « **Reply #14 on:** July 08, 2011, 02:42:32 pm »

EAC has an option to "Compare WAVs...". It's in the Tools menu. You can use that to compare the rip from J River to the rip from iTunes. "Compare WAVs" takes into account account the possibility of differing offsets when comparing WAVs. If the only difference is offset then it will say something to the effect of "missing samples". Which just means the differences are missing samples at the beginning or end (differing offsets). If there are different samples in the middle of the files it will tell you that.

My experience with ripping in J River (using secure mode) is that it does a very good job. Errors don't slip by. Differences compared to other rippers are due to differing offsets and differences in how pre-gaps are handled. If there's different samples (errors) between an iTunes rip and J River rip I'd believe the J River rip to be much more likely correct.

chew.isaac · « **Reply #15 on:** July 10, 2011, 06:20:57 pm »

Quote from: Vincent Kars on July 08, 2011, 03:12:09 am

I' not familiar with iTunes so the usual grain of salt applies.
iTunes uses QuickTime.
QuickTime can use WASAPI but as far as I know only in shared mode.
As a consequence all audio goes through the Win mixer.
The Win mixer dithers the signal.
As you are playing 16 bits, this might be audible

Beside the QT settings, do check the Win audio settings.
They should be equal in bit depth and sample rate.
If you play Redbook do check that both are set to 16/44.1

You might try to compare:
iTunes and MC using DS (Direct sound)
MC using DS and using WASAPI (exclusive mode)
It is possible that the differences are simply due to the driver used (DS or WASAPI).

But I must admit, there are a lot of people who do think MC simply sounds best.
An example: http://www.soundtrackforum.net/viewtopic.php?f=4&t=325

Thanks for the suggestions. I know what you mean. Will go try out later when I get home from work.

chew.isaac · « **Reply #16 on:** July 10, 2011, 06:29:28 pm »

Quote from: Frobozz on July 08, 2011, 02:42:32 pm

EAC has an option to "Compare WAVs...". It's in the Tools menu. You can use that to compare the rip from J River to the rip from iTunes. "Compare WAVs" takes into account account the possibility of differing offsets when comparing WAVs. If the only difference is offset then it will say something to the effect of "missing samples". Which just means the differences are missing samples at the beginning or end (differing offsets). If there are different samples in the middle of the files it will tell you that.

My experience with ripping in J River (using secure mode) is that it does a very good job. Errors don't slip by. Differences compared to other rippers are due to differing offsets and differences in how pre-gaps are handled. If there's different samples (errors) between an iTunes rip and J River rip I'd believe the J River rip to be much more likely correct.

I did try to use EAC's wav compare option. Problem is I don't understand the results. Comparing two wav of the same song ripped by eac vs itunes/MC, it says that song ripped by itunes and MC has Error type: 667 repeated samples, position 0:00:00.032. No error for the rip by eac. There is no other information. I don't know how to interpret that. Does anyone have a clue?

Alex B · « **Reply #17 on:** July 10, 2011, 07:03:06 pm »

Quote from: chew.isaac on July 07, 2011, 01:21:52 am

. I see differences in the data. Not just an offset. I just went to verify again...

A small offset difference can make the approximate graphical representation of the waveform in audio editors look different. You would need to zoom in to the sample level and actually align the files by cutting and pasting individual samples to make the waveform representations match in a zoomed out view. For example, if EAC's offset correction is +30, you would need to add 30 silent samples to the start and cut 30 samples from the end to make the file match the uncorrected files.

Quote from: chew.isaac on July 10, 2011, 06:29:28 pm

I did try to use EAC's wav compare option. Problem is I don't understand the results. Comparing two wav of the same song ripped by eac vs itunes/MC, it says that song ripped by itunes and MC has Error type: 667 repeated samples, position 0:00:00.032. No error for the rip by eac. There is no other information. I don't know how to interpret that. Does anyone have a clue?

I have never used EAC's wave compare tool and I don't know how it works, but you can simply temporarily set EAC's offset correction adjustment to zero to get the same audio content alignment as MC and iTunes create. You must use the same drive for ripping because different drives have different offsets. EAC's offset correction was invented to make rips from different drives have identical alignments, not making better rips. The Red Book audio CD standard allows offset variation in the pressing process and it is also normal that different drives have different read offsets. In any case these variations are quite insignificant. For instance, an offset correction of 441 samples is one millisecond. My Samsung ang LG drives have an offset correction of 6 samples in EAC.

If the files still differ after setting EAC's offset correction to zero, then probably at least one of the rippers has produced ripping errors.

You can use foobar2000's Binary Comparator tool for comparing the files easily (it's an optional foobar2000 component).

Frobozz · « **Reply #18 on:** July 10, 2011, 07:44:44 pm »

Quote from: chew.isaac on July 10, 2011, 06:29:28 pm

I did try to use EAC's wav compare option. Problem is I don't understand the results. Comparing two wav of the same song ripped by eac vs itunes/MC, it says that song ripped by itunes and MC has Error type: 667 repeated samples, position 0:00:00.032. No error for the rip by eac. There is no other information. I don't know how to interpret that. Does anyone have a clue?

The difference is at the very beginning of the tack. Almost certainly due to a difference in offset or gap calculation between the two rippers. If you did the comparison the other way you'd probably get 667 missing samples. The repeated or missing samples are most likely silence anyways.

I'd consider the iTunes rip and MC rip to be identical. The difference is only in offset. If you had differences in the middle of the file then it would be a case that one riper let an error go through.

Differences in offsets are no big deal. When you join the tracks back together you end up with the same stream of data for the CD as a whole. Just that the exact track boundaries will be offset a fixed amount between each ripper. Home stereo CD players play back with different offsets as well.

Alex B · « **Reply #19 on:** July 10, 2011, 07:47:27 pm »

I forgot to comment this:

Quote from: chew.isaac on July 06, 2011, 08:29:24 pm

By offset is it like the one wave starts playing the music slightly later than another wave? All 3 wav files that I have are in sync (pardon my non technical vocab).

An audio CD has a single continuous audio track. When the drives' read offset differ the entire CD audio track is shifted slightly. The individual tracks exist only in the separate table of contents. The ripped separate track files are "cut" from the continuous CD audio track during ripping according to the table of contents. All ripped tracks have the same amount of shift and thus for example gapless playback is not affected by the drives' read offset differences.

chew.isaac · « **Reply #20 on:** July 11, 2011, 04:32:25 am »

Thanks for all your sharing. I will go try them out. I will post the full results soon.

chew.isaac · « **Reply #21 on:** July 11, 2011, 09:58:41 am »

Thanks to all the suggestions.

I switch off the accuraterip in EAC and realized by analyzing my drive, it adds an offset of 667. By ripping using EAC without this offset, my wavform has no difference from the wavform by itunes and MC. I use both wav compare of eac and bit compare utility of foobar as suggested to verify that the wavforms are identical. So an offset does give the impression that there are extra data. This is probably because I don't zoom into the wavform (Alex B mentioned this), and 667 is enough offset to create an impression of data difference without a visible shift in the wavform.

So all 3 programs EAC, JRiver, itunes produce the same wav file for a good disc. All my discs are mine so they are all good. As I am ripping to wav, it is best to use J river for ease of auto-tagging. Itunes though rips the fastest (a lot faster), but then again some will argue about itunes secure mode.

Cheers!

glynor · « **Reply #22 on:** July 11, 2011, 10:05:03 am »

Good investigative work. Thanks for the update.

Matt · « **Reply #23 on:** July 11, 2011, 10:43:33 am »

Secure ripping is slower.

iTunes doesn't have a real secure ripping mode.

If you want to trade speed for security in Media Center, select "Normal" ripping mode instead of "Secure" mode.

chew.isaac · « **Reply #24 on:** July 11, 2011, 09:41:14 pm »

Quote from: Matt on July 11, 2011, 10:43:33 am

Secure ripping is slower.

iTunes doesn't have a real secure ripping mode.

If you want to trade speed for security in Media Center, select "Normal" ripping mode instead of "Secure" mode.

I see. Got it.

INTERACT FORUM

Author Topic: rip wav file compared to eac and itune (Read 10955 times)