Forum: help

RE: Win32 executable [ Reply ] By: Kaustuv Chatterjee on 2007-02-24 15:13	[forum:2779]
Dear Benjamin, The new release works perfectly and is a great help. Thanx, Kaustuv

RE: Win32 executable [ Reply ] By: Benjamin Roy on 2007-02-19 09:30	[forum:2758]
Dear Kaustuv, A new binary release is available, we have changed the read/write mode of the book files for Windows compatibility, so it should work find now. We will appreciate your feedback on this release. Regards, Benjamin.

RE: Win32 executable [ Reply ]
By: Benjamin Roy on 2007-02-15 17:44

[forum:2738]

Dear Kaustuv,

The email adress was correct and I received your files, I have reproduced the same bug as yours.

Well we have searched and find that this bug is a plateform dependant issue due to the binary files formats wich differ between UNIX and Win32 platform.

We will fix this bug as soon as possible and upload a new binary release for windows.

Thanks for reporting this bug and for your interest in Matching Pursuit Tool Kit!

Regards,

Benjamin.

RE: Win32 executable [ Reply ] By: Kaustuv Chatterjee on 2007-02-14 15:05	[forum:2734]
Dear Benjamin, I have forwarded the files to benjamin.roy@irisa.fr. I do hope that the email id is right. Regards, Kaustuv

RE: Win32 executable [ Reply ]
By: Benjamin Roy on 2007-02-13 12:57

[forum:2733]

Dear Kaustuv,

For the Warning related to the difference in number of sample, there is a previous thread that deals with this subject:
https://gforge.inria.fr/forum/forum.php?thread_id=2246&forum_id=109

The second Warning is more annoying, because it apparently means that the book isn't well formed. So, could I ask you to send me the related files (dict: test.xml, signal: s1.wav and book: tempbook.bin) in order to check what's happen exactly in this case?

Thanks in advance.

Regards.

Benjamin

RE: Win32 executable [ Reply ]
By: Kaustuv Chatterjee on 2007-02-12 17:50

[forum:2732]

Dear Benjamin,

Thanx for the response. I tried out mptk again, this time from command line with a different wav file and dictionary file. Again, mpd executed like a song without any errors. I used the following command line:

mpd -D c:\mptk\dict\test.xml -n 1000 -R 100 s1.wav tempbook.bin

then ran mpr:

mpr tempbook.bin temprecon.wav

and got the following errors:

mptk WARNING -- MP_Book_c::load<fid> - The book object contains a number of samples [9984] different from the one read in the stream [12544].
mptk WARNING -- MP_Book_c::load<fid> - could not find the </book> tag. <22 atoms added, 100 atoms expected.>
mpr msg -- The resulting signal has [9984] samples in [1] channels, with sample rate [12500] Hz.

This makes me think that the problem is not with the LAM matlab interface. Incidentally, I also tried out the bookread.m in the extras\matlab directory as suggested by you and got the same results. Also, the reconstructed wave i.e. temprecon.wav is a close approximation of the original wave (s1.wav) when plotted indicating that there are no issues with mpr atleast as far as the 22 atoms which have been correctly read by mpr in reconstructing the wave is concerned.

Looking forward to your response. Have a nice day.

Regards

Kaustuv

RE: Win32 executable [ Reply ]
By: Benjamin Roy on 2007-02-12 16:30

[forum:2731]

Dear Kaustuv,

I 've sent the description of this issue with the Mat2MPTKdemo from LAM matlab interface to the peoples involved in MPTK dev at LAM.

In first approach, Mat2MPTKdemo from LAM matlab interface seems to have compatibity issue with the later versions of MPTK (eg > 0.5.0).

Did you try to use the bookread.m script file located in the extras/matlab directory of the MPTK-Windows-bin-0-5-3-beta release ?

In fact, most of the MPTK package is fairly stable now. However, we haven't reached version 1.0 yet. In particular, the data format of the "books" is bound to change in the upcoming versions. We are working on a system of backwards compatibility, but it is not ready yet: in the meantime, the formats may change with each new version. We are conscious that this is a nuisance, and we are working on it.

I'll give you more informations about this particular issue asap...

Thank you for your interest in The Matching Pursuit Tool Kit!

Regards,

Benjamin.

RE: Win32 executable [ Reply ]
By: Kaustuv Chatterjee on 2007-02-10 17:59

[forum:2730]

Dear Benjamin,

Firstly, let me thank you for the amazingly prompt response to my request. To introduce myself, I am a doctor and have just recently taken to exploring the possibilities of decomposing EEGs using matching pursuit. To put it on record, I am a novice both in the fields of signal processing as well as programming (I do have a working knowledge of Matlab, though).

I downloaded the windows version of MPTK and the first hurdle I ran into was the non-availability of the C runtime library msvcr71d.dll. Once that was in place mpd seems to run perfectly. However, I am having problems with reading the books. On using the Mat2MPTKdemo of the LAM matlab interface, I get the following error after bookread has iterated to the 45th atom:

??? Error using ==> bookread
[] is an unknown atom type
Error in ==> bookplot at 48
book = bookread(book);

The dictionary used is the default.xml
Where am I going wrong?

looking forward to your help and thanx once again
regards,

Kaustuv

RE: Win32 executable [ Reply ] By: Benjamin Roy on 2007-02-09 17:26	[forum:2726]
We just release a Windows precompiled beta version 0.5.3. It's in the Files section in MPTK_win package. We expect your feedback for possible issue on this release. Thank you for your interest in The Matching Pursuit Tool Kit !

Win32 executable [ Reply ] By: Kaustuv Chatterjee on 2007-02-08 17:53	[forum:2713]
Is there a win32 executable of MPTK avaialable?