I have decided to transcode my collection from FLAC to WavPack. I have previously used FLAC because of it's fast decoding, fast sample accurate seeking and since i liked that it was open-source, but since wavPack also has the same advantages, and actually decodes faster than FLAC in -f mode while still encoding nearly twice as fast, then i have decided to change. Another reason is that bryant seems like a really nice guy, and really seems to care about his users... Anyway, i have a couple of questions that i would be happy if someone could help me with...
I have thought a little about if i should change from ripping multiple tracks to images instead... If i decide to change, then i will rip with EAC and encode + embed the Cue Sheet + log file with fb2k. I know how to set the diskwriter + CLIencoder up to encode + embed, if the image file + Cue Sheet is named CDImage.wav/CDImage.cue, and if i rename the log file to album.log :
-h -m -w "cuesheet=@cdimage.cue" -w "log=@album.log" - %d
I would like to have the images named after the album, and also not have to rename the log file so i thought about widcards... I have read on the WavPack site, that WavPack accepts wildcards... I would like to know if it also applyes for tagging ? In fb2k CLIencoder i tryed to encode an image file with this command line and it didn't work :
-h -m -w "cuesheet=@*.cue" -w "log=@*.log" - %d
I have decided to encode with -h, since it's fast enough for me, and still saves some space... Also which is most correct : -h -m or -hm (i think that they both work, but i want to use the most correct one). Thank's in advance...

-Martin.

Hi Martin,

First of all, thanks for considering WavPack!

Unfortunately, WavPack will not (as you found) accept wildcards in the tag file fields. It probably would not be that hard to put in; I'll consider it for 4.3 (of course, it will have to abort with an error if more than one file match the wildcard).

There's no better or worse way to combine the commands. Having them separate makes them easier to understand at first, and then you can put them together to save typing later. I use lines like -hxb256nycm all the time.

BTW, remember that files encoded with -h decode about half as fast as FLAC. If decode speed is very important, you can get a good portion of the extra compression of the high mode without the slowdown on decode by using -x instead of -h, but the encode will take much longer.

David

Nice idea. Could be very useful for transcoding. If it's easy/bloat-free to implement it's got my vote.

I only discovered recently that Diskwriter (or Converter) won't process TAGZ in the command line. Otherwise, until WavPack 4.3 (if implemented), you could have used something like:



With regard to using "@*.<ext>": Wouldn't that potentially, under normal/current conventions, look for a file in the same folder as WavPack? I guess it would have to be implemented so that the path to the source would be used as the folder to search? Something to consider (as if you haven't already!).


I agree. One of the various reasons I intend to switch to WavPack (when I get my external hard drive which should be in the next month or so) is David's active participation and respose to users' requests . That said, I hope you're not insinuating that Josh is an uncaring b*stard?

Hello David Thank you so much for answering my questions, i really appreciate it Also thank you very much for considering to add wildcard support for tags, that would be great From now on i will only use WavPack - Thank's again David


Hello Synthetic Soul In fb2k then the files source path is used, so theres no problem... I used this line when i had wavpack.exe and my image+cue+log files in two different directories : -h -m -w "cuesheet=@cdimage.cue" -w "log=@album.log" - %d

I am sorry about the way i wrote my first post That was absolutelly not what i meant... I have great respect for Josh and i am sorry about my bad wording

-Martin.

Hi Martin,

Ah, OK. My misunderstanding. Thanks for the info. I don't think I thought my statement through at all - obviously the normal/current convention is actually to look in the current folder, whether that be the folder that contains WavPack, the source file, or any other.

I'm sorry Martin, I was just joking. Josh is also an active forum member, and very well respected. I was just causing trouble by chosing to read something between the lines that wasn't there. I will stick my neck out and say David seems more open to suggestion than Josh (the FLAC CUESHEET block being an example), but both applications and authors have my respect and support. According to the latest poll FLAC is still by far the most popular lossless format, so Josh must be doing something right. I suspect WavPack's share has increased since that poll.

Thank's Synthetic Soul, that makes me feel better The last thing i want to do, is to offend people that work hard at coding great software and then gives it away for free to the community

-Martin.

Just for the record, both Josh and Matt are nice guys too...

Of course, both FLAC and Monkey's Audio are far more popular than WavPack and so both of them can "rest on their [well deserved] laurels" if they like. Obviously, the only way WavPack can ever hope to match them is by responding to the requests of users (and maybe providing unique features).

The irony here is that WavPack is older than both of the other programs, and it was my past resistance to user's requests that probably sealed WavPack's fate. Matt likes to joke that he created Monkey's Audio because I never created a winamp plugin for WavPack, and I think Josh created FLAC because Matt would not open the source of Monkey's Audio. And some members will remember the ridiculously slow seeking of WavPack files before version 4.0 arrived!

Anyway, I deeply appreciate the support of the entire HA community in making WavPack what it is today. Thanks!

It's very interesting to hear that brief overview of the top three lossless codecs. I was aware that WavPack has been around a while, but not that the development of these codecs was quite a linear process. It's also interesting that your willingness to listen to users hasn't always been so evident - I hope this thread is another small pointer (I'm sure there have been others) that your current strategy is working well for WavPack.

FYI: I don't really have much interest in the hybrid functionality of WavPack, although I wouldn't dismiss it as unusable. My main interest in the codec, in order of priority, are:

• Error tolerance
• Development appears to still be moving fast and well
• Support for multi-format WAVE (inc. multichannel)
• The @ file reading/built-in pipe support
• Open source and cross platform (even though I'm Windows through and through).

I only recently realised that those extra few GB I would save using Monkey's Audio are actually negligable considering the factors above.

The error tolerance of my APE files concerns me. When I shift my DVD backups to hard drive they will be converted to WavPack first. All future archiving will then be WavPack (-hm) only.

Thanks for the option. Good luck for the future.

Hello David I know that you have said that it doesen't matter, and i understand that, but could you please tell me which one of these command lines you would use :

1: -hm -w "artist=%a" -w "title=%t" -w "album=%g" -w "year=%y" -w "track=%n" -w "genre=%m" %s

2: -hmw "artist=%a" -w "title=%t" -w "album=%g" -w "year=%y" -w "track=%n" -w "genre=%m" %s

(Only the red part is different...)

I have decided that i will continue to rip to multiple tracks, and not to images afterall, but i just want to be using the most "correct" WavPack command line...

On Case's EAC configuration page : http://www.saunalahti.fi/cse/EAC, there is listed this command line under WavPack :

-m -w "artist=%a" -w "title=%t" -w "album=%g" -w "year=%y" -w "track=%n" -w "genre=%m" %s

On that page there are three other codecs listed which uses APEv2 tags(with wapet), but they all have the tag fields defined with the first letter in uppercase.

Monkey's Audio : %d -t "Artist=%a" -t "Title=%t" -t "Album=%g" -t "Year=%y" -t "Track=%n" -t "Genre=%m" mac.exe %s %d -c2000

OptimFROG : %d -t "Artist=%a" -t "Title=%t" -t "Album=%g" -t "Year=%y" -t "Track=%n" -t "Genre=%m" ofr.exe --encode %s --output %d

LAME : %d -t "Artist=%a" -t "Title=%t" -t "Album=%g" -t "Year=%y" -t "Track=%n" -t "Genre=%m" lame.exe --alt-preset standard %s %d

Do you know why the WavPack example has all lowercase tag fields in contrast to the three other codecs ? And what would you recommend that i use : Artist or artist ?

Thank's in advance.

-Martin.

I echo Martins request,i also am in the middle of archiving all my music into
wavpack but have paused after seeing several threads open,now i'm confused.
I'm not concerned with hybrid(not doggin it)just don't need it.
I want the killer command line i can punch into EAC and not be worried
I just purchase a new Lite-on drive got the latest EAC and wavpack and
some White Stripe CD's arriving in the mail.
I need the hook-up.

The APEv2 tag fields in the HA wiki are in title case, though I don't think it makes any difference. I use those command line options (with -mhx6 instead of -h) and they work fine for me.

afaik, they will be bit identical. test it out if your are skeptical.

I'm not skeptical... I just asked David which line he would use of the two...

-Martin.

Okay, let me try to answer everything without making it more complicated...

Martin, I would use your #1 line because I like the -w separate so it's easy to see clearly. But they'll both work. As for the caps in the tag field names, I like the first letter capital. The definition of APEv2 tags says that tag readers should ignore case, so it really should not matter. I don't think Case made the command lines different on purpose.

Tinyvillager, here is the simplest, basic EAC command line that will work with WavPack and put in APEv2 tags:



You can read the WavPack documentation and decide if you want to add any other options (a lot of people add -h for high compression mode), but you don't have to.

In that other thread I created a complicated command line so that people could select some WavPack options using the selection buttons on that EAC compression window, but you don't need that if you just want the basic mode.

Thanks bryant
That's actually the command line i've been using,thank god,
cause i've already archived like 15 cd's.Just wanted to
clear it up and put my paranoia to rest.Keep up the good work.


P.S.
Just googled paranoia,(can't spell)and the first site to pop up
was www.xiph.org/paranoia/

Google is scary

I guess the following could be your "killer" wavpack EAC command line. It tags all the information EAC passes over (including %x, %e, %f, %b). By selecting "High quality" in the EAC compression options dialog, the -h switch is enabled:

David, thank you so much for your continued support, i really appreciate it


David, i noticed that you are using the %d destination filename variabel in your commandline. The WavPack commandline on Case's configuration site dosen't :

Would you recommend to use the %d or not ?

Thank's in advance.

-Martin.

Hi Martin.

I use the %d file specification only because it's required for the "hack" that I put into wavpack.exe to allow the correction files to automatically work with EAC. For that hack to work, the third filename is the name for the correction file, but if you don't specify the destination filename then you can't have three filenames.

If you're not using the hybrid lossless mode then you can get away without the destination filename, but I still use it.

Thank's David Could you please explain to me what exactly it means when you say that tag readers should ignore case. Does it mean that if i for example has a field called Artist, then the tag reader will display it as ARTIST(if the program is using all uppercase as standard for the displayed tags)... Or does it mean that if the tag reader is looking for a field called Artist, then it will find it in the file no matter if its called ARTIST or artist...

Thank's in advance.

-Martin.

The latter. Tag readers should find the tag in the file no matter what case is used, even "ArTisT".

And tag readers can display the tags any way they like. For example, foobar2000 displays "ARTIST" no matter what the case is in the tag. In fact, foobar2000 will even display the field "Year" as "DATE" and "Track" as "TRACKNUMBER"...

David, thank you so much for all the help you have given me

-Martin.

Thanks for all the info!

I've got a few questions regarding converting audio files to WavPack (v4.2) via foobar2000 (v0.8.3)...

1 - I'm using a very simple string in the Parameters box in foobar's CLI settings which seems to work fine:Is this sufficient for general encoding, or is there some room for improvement?

2 - Is there any benefit or drawback to enabling the "Encoder requires accurate length" option? (I've never quite understood what this option is for.)

TIA!

~esa

If I've understood this right you should either enable the "Encoder requires accurate length" option or use the -i switch with wavpack in fb2k.

Personally I use -himt - %d (high compression, ignore length in wav header, make md5, copy timestamp).

Thanks for the help, Mr_Rabid_Teddybear..!

I've tried this myself in 0.8.3 and 0.9 and I can't get it to work unless I put in the full path to the cuesheet.

Any pointers?

The command I'm using is:


If I use WAVPACK_DEBUG.EXE the report states:

...
8: -w
9: CUESHEET=@CDImage.wv.cue
10: -
11: C:\DOS\testing\Kings of Leon\CDImage.wv
error in tag spec: CUESHEET=@CDImage.wv.cue !

As I say, it works fine if I use an absolute reference to the cuesheet.

Edit: With further testing it seems it will work if the cuesheet is in the same folder as WavPack. This makes me suspicious that foobar is actually running from the WavPack folder, and not the source file folder, as first suspected.

Can anyone confirm or deny?

To test it out, i have just ripped a CD to an image + Cue Sheet and then encoded it in fb2k with this line :

-hm -w "Cuesheet=@cdimage.cue" -w "Log=@album.log" - %d

I had cdimage.wav + cdimage.cue + album.log in D:\Temp, and wavpack.exe in C:\Programmer\foobar2000.

-Martin.

Hi Synthetic Soul & Martin H

I actually made some progress on implementing this metadata from file stuff that was discussed above, and it's already in the special version I did with the experimental noise shaping. I didn't mention it then because it was very preliminary, but it does seem to work (and if you're doing lossless that version should be identical to the standard).

There are two enhancements. First, it accepts wildcards in the file spec as long as there is only one match. So you can use "CUESHEET=@*.cue" for instance, as long as it's the only cuesheet in there. If it can't find the file in the directory that WavPack is run in, it then also tries the directory that the source file is in.

However, after reading this I realize that this won't work in all cases. Perhaps I should try the destination directory also if the input spec is stdin?

Thank you very much for spending the time to test and confirm Martin, and for responding to my PM.

I have tried responding this morning about five times now - coming to different conclusions each time - and wasting the last hour.

Before I post my confused musings can I ask what output file name format you are using please Martin? Edit: Also, what version of foobar?

You may be pleased to know that I have actually had some success this morning - but not all the time! Update: I am having no luck at the moment with any settings, so the ratio has fallen to something like 3 successes to 30 fails...

I think this is to do with using STDIN as the input (source), but I need to do a load more testing to make any clear conclusions. I have made many so far this morning - some I've disproved and others I'm just not sure about. Update: I'm having no luck at the moment using %s either, so I'm not so sure.

I think this will be useful. My vague suspicion at the moment is that foobar is using the source directory as the current directory, but because STDIN is being used it is using a "cached" source directory... which is obviously not ideal. As I say, this is only a suspicion at the moment. My head is already spinning from this morning's tests. Update: I don't think this is the reason as using %s is proving unsuccessful for me too at the moment.

To allow users to use STDIN while embedding from a file it would be useful to check the destination directory also. I know the WAVPACK directory is the default place to check, but in normal use the file, if a relative path, is normally going to be in the source or destination folder - and if there's no physical source...

I'm not sure if I have the "special version" or not! I'll download the latest version from wavpack.com - but I'm keen to diagnose my current setup, as I know it will and won't work - and would like to find out the circumstances for each. Update: I found the "special" version in this post. I have tested using this (@*.cue), with %s instead of stdin, and it works. I'm still detirmed to work out the varying and frustrating results I'm getting with the "normal" version though. Although I managed to get it to work this morning subsequent tests have all (using %s and stdin) been unsuccessful! NB: I realise that this isn't a WavPack issue. It's me or foobar that's confusing me. Thanks for the * code though, as that may end up being the way I have to go.


Thank you both.

WAVPACK.EXE : 4.2 2005-04-02 (EXE modified date 2005-04-02 12:00).

WAVPACK_DEBUG.EXE : 4.2x 2005-07-13 (EXE modified date 2005-07-13 23:37).

WAVPACK.EXE (special) : 4.x 2005-09-13 (EXE modified 13 September 2005 20:45:20)

OK, I think I may have it. (don't give it to me!)

This is why I thought foobar was "caching" the source directory.

foobar appears to use the folder last specified when the output folder is specified by the user manually via dialogue ("Convert > Convert to..." in 0.9 or "Always ask before writing" checked in 0.8.3).

So, if you select the "Same as source file directory" checkbox in 0.8.3, or use "Convert > Convert to same directory" in 0.9, the folder which you last specified via dialogue box is used as the current directory - not this source file's directory.

To confirm my final conclusion I created the following:

• C:\One\test.cue
• C:\One\test-one.wav
• C:\Two\test.cue
• C:\Two\test-two.wav

The last folder I had specified via dialogue before this test was C:\DOS\Testing

1. I go to convert to same directory (Convert to same directory) on test-one.wav. Fail.
2. I go to convert to same directory (Convert to same directory) on test-two.wav. Fail.
3. I go to convert test-one.wav and specify to output to C:\One (Convert to...). Success!
4. I go to convert to same directory (Convert to same directory) on test-one.wav. Success!
5. I go to convert to same directory (Convert to same directory) on test-two.wav. Success! However the cuesheet embedded is C:\One\test.cue - not C:\Two\test.cue (as C:\One is the last folder I specified manually by dialogue).

The reason I was getting so many fails in my test was presumably as the last folder I specified by dialogue did not have my test cuesheet in, and I wasn't specifying the output dir each time to save me time while testing!

So, to conclude: works fine if you specify the output directory manually - but not if you select to automatically output to the same directory as the source. Edit: works fine if you specify the output folder manually, only do one file at a time, and use 0.9.

NB: This isn't brilliant for me as I want to do this en masse - which means that outputting to the source directory is the easiest way for me. I guess I'll use the @*.cue method with 4.x. I just tested by converting both test-one.wav and test-two.wav, in one run (in 0.9), and manually specified the output folder as C:\. Both files converted, but both files have C:\One\test.cue embedded. Make of that what you will! (remember C:\ has no test.cue file)

Edit: Actually this may well only apply to 0.9. I can't get it to work in 0.8.3 at all (Edit: unless I put test.cue in the same folder as WavPack).

Martin, please tell me you are using 0.9!

Also, perhaps you could try converting two+ files at a time, take a look at the embedded cuesheets, and confirm that they are both/all the same one.

Hi David Thank you very much for implementing that feature. I had thought a little about if i should begin to make images instead of individual tracks, but i have decided that i like the individual tracks approach better, but to all the people that make images then this new feature is really great... Thank's again


Hi Synthetic Soul Sorry that i forgot to tell you what version i was using. I am using v0.8.3... To test it out for you, i have just ripped 2 CD's to images, and encoded them in fb2k... First i encoded them with the option 'Same as source file directory' enabled, and after that i encoded them again, but this time specified an output directory. Both the times, the two albums had the same Cue Sheet + Log file embedded

-Martin.

Thanks again Martin.

I'm not sure how you can get it to work in 0.8.3 and I can't, but I've spent enough time on this already.

As you have also noted that only one directory is being referenced for the cuesheet and log then the process is not robust enough for me to use as is.*

I will use the special version 4.x specify @*.cue, and use %s instead of STDIN.

Either that, or I may just write a batch file to undertake the process. It is likely I will lose my ARTIST and ALBUM APEv2 tags -unless I can find a way to export to text file and then re-apply (Tag may do this) - but I am thinking that as long as I have the cuesheet I don't really care, and a batch file to do the whole process would be a lot easier, and quicker in the long run.

Thanks again for your help.

Edit:

* FYI My purpose is to convert around 350 APE files with embedded cuesheets and ARTIST/ALBUM APEv2 tags to WavPack. I am transcoding from DVD storage, so I will have to work a number at a time. With a btach file i should really be able to do the whole process, so I can just stick a new DVD in, double click the bacth file, and eventually I will have my WavPack files with embedded cuesheets (and possibly APEv2 tags) on hard drive.

Edit 2: Cool, I can use Tag, something like:

TAG.EXE --fromfile %1 "G:\CDBackup%~pn1.wv"

... should copy the tags directly from the APE file to the WV file.

@Synthetic Soul:

You are welcome I hope that you can get the job done without to much trouble...

-Martin.

Thanks Martin.

I actually wrote the batch file last night and had a brief test with disc #1. It seemed to work very well.

This is how it works:

1. Iterate through APE files on disc
   1. Duplicate APE file's folder structure on local hard drive (MKDIR)
   2. Convert APE file to WAV on local hard drive (MAC)
   3. Convert WAV to WV (WAVPACK)
   4. Delete WAV
   5. Copy tags from APE to WV (TAG)
   6. Copy CUE to local hard drive and change FILE reference to <name>.wv (SED)
   7. Embed cuesheet in WV (TAG)
2. Iterate through LOG files on disc
   1. Copy file from disc to hard drive, maintaining folder structure (COPY)

I'm very pleased with this as I can just stick a disc in, double-click my batch file, and away it goes.

Once I'm completely happy with it I'll comment it up and post it on my site. I know of at least one other member that may find it useful.

Edit:

The current version actually uses the version of MAC with pipe support available here, and can be donwloaded from my website here. This changes the process to:

1. Iterate through APE files on disc
   1. Duplicate APE file's folder structure on local hard drive (MKDIR)
   2. Pipe data from APE file to WV file (MAC + WAVPACK)
   3. Copy tags from APE to WV (TAG)
   4. Copy CUE to local hard drive and change FILE reference to <name>.wv (SED)
   5. Embed cuesheet in WV (TAG)
2. Iterate through LOG files on disc
   1. Copy file from disc to hard drive, maintaining folder structure (COPY)

@Synthetic Soul:

Great to hear that it's working out for you It sounds like a really great batch script you have made, and i'm sure that many people would find that extremely valuable when they are up to do a task like yours... Well done

-Martin.

hi bryant i have been trying out wavpack recently and was wondering if you planed to inculde a frontend at some point? if you dont want to have to write your own maybe you could ask speek if you could use his as its not bad at all link