Ripping software and settings

Hardware and software, modifications and DIY

Moderator: Staff

Post Reply
hcl
Very active member
Very active member
Posts: 360
Joined: 2008-01-13 11:03
Location: Göteborg
Contact:

Post by hcl »

jrob wrote:... I have also compared the two resulting (uncompressed) flac files in foobar 2000 using the bit comparator plugin and there are differences although they are very small. ...
Did you get any error report while ripping the discs?

As I understand it, differences implies bit errors and small differences implies that there has been some interpolation to reduce the error. Interpolation require that the error is detected thus there should be an error report.
jrob
New member
New member
Posts: 4
Joined: 2007-06-12 17:52
Location: London

Post by jrob »

Hi hcl,
I cannot absolutely confirm either EAC or ripstation are precisely extracting every bit in the correct order however, EAC does verify the rip as accurate by checking with the online accurip service and reports every track as 100% quality which I think means no error correction was required. I have also ripped the same CD 5 times with EAC and always got exactly the same results every time. Perhaps somewhat worryingly with ripstation there were differences reported in 3 of the 5 rips using Foobar bit comparator! This subject is being discussed in more detail on the slim devices forums
http://forums.slimdevices.com/showthread.php?t=44534 .

To ensure there are no problems with my computer hardware or software (it’s a newish quad core P4 with a Sony DVD RW drive) I will repeat the same tests using the same CD on a couple of other PCs with different DVD drives over the weekend.
Regards,
jrob
Nicolav
Active member
Active member
Posts: 201
Joined: 2008-04-07 22:46
Location: Italy - Rome

Post by Nicolav »

My experience is absolutely subjective, anyway this is my conclusion:

To ripping CD I have tryed EAC, ripstation and dbpoweramp.
EAC in my opinion is the best in terms of tune, the second best options seem to be ripstation and dbpoweramp as third...
Both ripstation and db seem to lack some life in to the music, much more less involving than EAC, I don't know why...
This result above regardless the drive used, Plextor PX-716A, NEC 3450 and the old Plextor 40X scsi. Anyway the best of these in tems of tune is the Plextor PX-716A.
As NAS I own a Qnap 109 with a Seagate barracuda 7200.10 500 Gb. I have found this much more musical than a pc with a 500 Gb Samsung hd. And the Qnap 109/Seagate 7200.10 sound better than Qnap 109/WD GP 750 GB owned by my friend Paolo.
Json
Member
Member
Posts: 13
Joined: 2008-01-26 23:05

Post by Json »

I have also made some, admittedly small but still, comparisons between two rippers/coders.
One was EAC on a PC, the other was Max on a Mac.

The resulting Flac files are different, the EAC files are a bit smaller.

When converting back to WAV both files were the exact same size however when comparing them the first result was that there were rather big differences.
However when I checked it more thoroughly I found that the only difference were that 24 NULL bytes were in different locations in the files.
In one of them the 24 NULL bytes were found before the music data and in the other they were after.

When taking that into account there were absolutely no difference at all between the files.

Listening to the two FLAC files I also found them to sound the same.

What about trying to "duplicate" the FLAC settings of EAC in the other rippers/coders (be it RipStation or Max etc.)? Shouldn't that produce the same FLAC file?
If so, all the RipStation (and Max) files could be re-encoded using the EAC FLAC settings and thus produce the exact same FLAC files. And if the FLAC and the resulting WAV are the same the sound should be the same. If not I would really like to see some explanation for the difference.

I saw somwhere that EAC supports "replay gain" but I don't know if it is enabled by default. If it is I suppose some changes could be found when comparing the sound with a ripper/encoder that does not set "replay gain" (supposing the DS actually cares about the "replay gain" setting).
User avatar
lejonklou
Administrator
Administrator
Posts: 6524
Joined: 2007-01-30 10:38
Location: Sweden
Contact:

Post by lejonklou »

Json wrote:And if the FLAC and the resulting WAV are the same the sound should be the same. If not I would really like to see some explanation for the difference.
Yes, two identical FLAC files really should sound identical. But please note that you may have two different FLAC files that, when converted to WAV, becomes identical. These two FLAC files (which contain the same information, but are encoded differently) can sound slightly different - probably depending on how the software and/or hardware in the DS is handling the file.

I am not saying this is of any importance, just making sure we agree on what is what.

Do you think it's possible that our current preference for EAC could change when the DS software is updated? If so, comparing ripping software becomes rather pointless...
Json
Member
Member
Posts: 13
Joined: 2008-01-26 23:05

Post by Json »

Hi Fredrik,

Yes I agree that if you have two Flac files that are different, there might be a chance that they sound different. I read in the Flac entry in Wikipedia that higher compression (i.e. resulting in smaller files) also means that un-packing requires more CPU-power. So in our example the EAC files would require more power from the DS for playback than the RipStation or Max files.

However when I compared playback of Flac files ripped/encoded with EAC with Flac files ripped/encoded with Max I didn't find any sound quality difference even though there were differences in the files. It might be that I simply can't hear it of course. :)

But what I wanted to "get at" is that IF there are differences in the Flac playback quality, wouldn't it be possible to get he exact same playback quality if a Flac file created by RipStation would be transcoded to Wav and then back to Flac using the same settings that EAC uses? (that is, to produce a file that is identical, possibly except for some header information, to the Flac file produced by EAC).
That is, it would not be necessary to re-rip the disc with EAC.

If there are differences in Flac playback quality I hope it could be rectified with new DS firmware, since I would consider such behaviour a "bug" (since Flac is basically Zip for audio).
However if the differences are due to some hardware phenomena (such as CPU power consumption or amount of data that fits in the buffers) it might be more difficult.
But in that case I hope Linn would specify which Flac settings they think produces the best results.
Json
Member
Member
Posts: 13
Joined: 2008-01-26 23:05

Post by Json »

Ok, I checked one more thing. I took one song, ripped with Max and encoded to Flac and did the following:
- First I converted it to WAV using Max.
- Then I converted it back to Flac using Max AND using EAC.

The two resulting Flac files differ in size by only 700 bytes and it is the EAC file that is the bigger one. When using RipStation or Max for ripping and Flac encoding the files will be bigger just because these two include album art, and EAC does not, but for this case the album art and tags were stripped away when converting to WAV. 700 bytes difference is not much for a 45 MByte+ file, which makes me conclude that the encoding settings of Max and EAC could very well be the same.

So to validate that I checked inside the files using a hex editor and I found that the beginning of the files differ as there seems to be some identification information that differs. And there is a block of NULL bytes that appear in one of the files but not the other.

But when the "music" portion of the file starts the files are identical. I cut away the header and NULL differences at the beginning of the file and did a byte compare of the rest (44 450 961 bytes of a total of 44 460 570 bytes for the file with headers and NULL) and there was not a single bit difference.

Then I opened the "full" EAC and MAX files in the "Tag" tag editor. I found that the MAX file had no tags whatsover (since they were stripped away when converting to WAV and not added when the file was converted back to Flac).

The EAC file on the other hand had the following tags:
ALBUM (empty)
ARTIST (empty)
DATE (empty)
GENRE A Capella
REPLAYGAIN_ALBUM_GAIN -9.12 dB
REPLAYGAIN_ALBUM_PEAK 0.97128296
REPLAYGAIN_REFERENCE_LOUDNESS 89.0 dB
REPLAYGAIN_TRACK_GAIN -9.12 dB
REPLAYGAIN_TRACK_PEAK 0.97128296
TITLE (empty)
TRACKNUMBER 00

Why EAC thought it is A Capella is beyond me....

However I found it quite interesting that the replay gain fields are set. If the DS acts on these, and changes playback volume because of it even if it is ever so slightly, could this not be one reason for hearing differences between EAC files (with replay gain) and non-EAC files?

I have not had the opportunity yet to listen to this exact track, so I can't comment on whether or not I can hear any differences.

So I suppose the next step is to remove the "--replay-gain" setting from EAC and rip some files and see (or, rather, hear) what happens.... but that is for another day.
User avatar
lejonklou
Administrator
Administrator
Posts: 6524
Joined: 2007-01-30 10:38
Location: Sweden
Contact:

Post by lejonklou »

I'm impressed, Json. Good analysis!

Those GAIN tags are interesting... Do they control anything regardless of what the rest of the file looks like? Or do they merely state facts about how the file is defined and made up?
But what I wanted to "get at" is that IF there are differences in the Flac playback quality, wouldn't it be possible to get he exact same playback quality if a Flac file created by RipStation would be transcoded to Wav and then back to Flac using the same settings that EAC uses?
Yes, I agree it should be possible.

I wish we had a FLAC guru here, telling us how this works in detail.
User avatar
vicdiaz
Active member
Active member
Posts: 248
Joined: 2007-02-06 04:37
Location: Trujillo Alto, PR U.S.A.
Contact:

Post by vicdiaz »

Since RipStation MicroDS support 'ReplayGain' it should be interesting in comparing values between RS and EAC.
Vic
Ivor's "Tune-Method Seminar" Alumni
Json
Member
Member
Posts: 13
Joined: 2008-01-26 23:05

Post by Json »

Thanks Fredrik!
I read about replay gain on wikipedia (article "replay gain") and they say "Typically, the gain value and the peak value are then stored in the audio file as metadata, allowing Replay Gain–compliant audio players to automatically attenuate (or in some cases amplify) the output so that such files will play back at similar loudness to one another."
There is more info there about how it works etc.

It is also possible to post process flac files to add replay gain tags, "metaflac" seems to be one such tool. I have not tried it.

And yes, I agree about the FLAC guru need.... :)

@Vic:
Yeah I agree that checking values etc. set by RipStation would be a good next step.
Do you know if there is a way to check what FLAC settings RipStation uses?
Pediatrik
Active member
Active member
Posts: 100
Joined: 2007-01-31 17:19
Location: Visby, Sweden

Post by Pediatrik »

Very interesting results indeed, Json!

Which settings do you use with Max?
paolo
Active member
Active member
Posts: 125
Joined: 2007-01-31 12:49
Location: Rome, Italy

Post by paolo »

Hi Json,
the "ReplayGain" matter did attract my attention too when I was doing some comparisons to find the best sounding settings for EAC.
In my experience it doesn't make any difference to sound quality. The reason why it is included in Linn EAC profile is not clear, infact I'm reasonably sure that the both K and Aku DS simply ignore its indications (they cannot adjust output volume infact) - though I'm not sure about Sneaky DS.

Paolo
User avatar
Music Lover
Very active member
Very active member
Posts: 1673
Joined: 2007-01-31 20:35
Location: In front of Lejonklou/JBL/Ofil

Post by Music Lover »

paolo wrote:I'm reasonably sure that the both K and Aku DS simply ignore its indications (they cannot adjust output volume infact)
Yes, without having any evidence it seems like DS is not adjusting the volume.
I have compared the flac files with the the DS output volume...
It's all about musical understanding!
Json
Member
Member
Posts: 13
Joined: 2008-01-26 23:05

Post by Json »

Pediatrik wrote:Very interesting results indeed, Json!

Which settings do you use with Max?
Hi Pediatrik,

I just used the default setting, which is compression level 5.
That also means that Max sets the following:
- Apodization tukey(0.5)
- Enable mid-size coding for stereo input.
- QLP coefficient precision 0
- Minimum residual partition order 0
- Maximum residual partition order 5
- Maximum LPC order 8

And no, I am sorry but I don't know what those values mean. :wink:
Pediatrik
Active member
Active member
Posts: 100
Joined: 2007-01-31 17:19
Location: Visby, Sweden

Post by Pediatrik »

Hi Json!

I guess this means that you use the Basic Ripper! Do you know the difference between this and Comparison Ripper/Cdparanoia?

Do you rip to a single file with a cue sheet or is that unnecessary when using LINN GUI?

Which tag mapping do you use, FLAC or Monkey's Audio?

Do you know how to get album art that isn't avaliable on Amazon?

Thanks!
Json
Member
Member
Posts: 13
Joined: 2008-01-26 23:05

Post by Json »

Hi Pediatrik!

No I am using the cdparanoia ripper. I don't think the flac settings really have anything to do with the ripper. The ripper rips to something non-compressed (wav maybe??) and then the flac encoder is started on the ripped file.

cdparanoia uses "re-reading" of sectors if it encounters an error. I think it is similar to what EAC does.

I have not tried the comparison ripper.

I rip to one file per song. Maybe I should have ripped to one file per album with cue-sheet, but.... I have quite a few CD:s left to rip, many of them classical music, and I thought I'd try to rip some of those to one file with cue sheet to avoid gaps, but I would really prefer to have one file per "song" even for those (if I can avoid gaps).

Maybe it is possible to "transcode" from one song per file to one album per file??

For discs that are not availble from amazon, mostly Swedish discs, I try all sorts of sources. Ginza.se have some, googling finds some, and I will have to scan album art myself for some albums.

I use media rage to put album art in the files when needed. I would prefer to use Tag, but it currently does not support album art.
Pediatrik
Active member
Active member
Posts: 100
Joined: 2007-01-31 17:19
Location: Visby, Sweden

Post by Pediatrik »

Json, now I've tried ripping a number of CD:s with Max using both Basic Ripper and Cdparanoia. Unfortunately the result is that Max shuts down after ripping the first track. :evil: Does it sound familiar or could it be a bug in my Max (0.8)?

Buy the way, this discussion is a bit off topic. Maybe it should continue somewhere else. Admin?!
User avatar
lejonklou
Administrator
Administrator
Posts: 6524
Joined: 2007-01-30 10:38
Location: Sweden
Contact:

Post by lejonklou »

Thanks Pediatrik.

This topic was split from "Best NAS for a DS"
Post Reply