search

mtpearce / idyom

http://mtpearce.github.io/idyom/
GNU General Public License v3.0
45 stars 9 forks source link

Database Management/Upload #42

Open ecm2021 opened 3 years ago

ecm2021 commented 3 years ago

Hello,

I've just downloaded IDyOM onto my laptop to use as part of my master's thesis, however after following the database management procedure as seen in the wiki section, IDyOM seems ubale to read my Kernscore file corpus. I'm rather new to emacs personally, but after going over the problem with my dissertation supervisor, she also seems unaware of the problem's source. I've attached a screenshot below. Thank you, Eden

IDyOM
Kappers commented 3 years ago

Hi there! When using idyom-db:import-data to import a folder of e.g. kern/MIDI files, you need to ensure the path specified ends with a slash - this indicates that the path is a folder, instead of a file. In your case, this would mean changing /home/eden/HanChinese to /home/eden/HanChinese/. Let us know if that fixes the problem!

ecm2021 commented 3 years ago

Hello, I'm sorry I never saw this earlier - yes, I managed to get it fixed thank you!

I do however have one different issue - I am trying to get the ioi-ratio from a corpus of midi files, however whenever I try to do so, I get an error saying "Lisp connection closed unexpectedly: connection broken by remote peer". This is particularly odd able to get the ioi-ratio for my Chinese corpus (this was a corpus of kernfiles however), and this Scottish corpus worked fine when measuring pitch. I've attached a screenshot below. Thank you.

Troubleshoot
Kappers commented 3 years ago

Hi, sorry for such a delayed response, were you able to resolve the problem?

It isn't clear exactly what is causing this error - whether it is related to issues in the data/IDyOM or just Emacs.

Please can you try running the same commands directly in SBCL and share the output? If it works, then this might related to your Emacs setup/configuration, and if not then it might provide a more detailed error report.

Thanks! Tom

ecm2021 commented 3 years ago

Hi Tom,

Sadly, I haven't been able to identify the root of the problem. Originally, my supervisor thought it may be an issue with the corpus, and decided to take out the piece that error message was pointing to. Now, however, it keeps coming up with the error surrounding duration. My supervisor did an analysis of the rhythm using Humdrum, and there shouldn't be any irregular rhythms within the corpus.

I also tried to analyse the corpus using SBCL, as you mentioned, but still there seems to be an error.

error

Picture2

Thank you, Eden

.

Kappers commented 3 years ago

Sorry to hear you are having trouble. It is hard to tell what the issue is at a glance, please can you provide a download link to the culprit melody? Or, describe how to find it online? When possible, I will take a look and try to understand what the problem is.

For now, I'd suggest you remove this melody from the dataset, and try to keep going with rest of the corpus.

Tom

mtpearce commented 3 years ago

When running lisp in the terminal, the error message shows that SBCL has run out of memory and closed which I suspect will be why Slime was telling you that the lisp connection had closed unexpectedly. (By the way, the inferior-lisp buffer in Emacs gives you the same view that you can see in the terminal.)

It's not clear why this is happening in this case but it sometimes happens with very large corpora and/or viewpoints with very large alphabets.

The backtrace in Emacs shown above shows an error a duration value of 3/64 into the database, when an integer value is expected. The default timebase in IDyOM specifies that a crotchet (quarter note) corresponds to 24 basic time units, which is sufficient for the majority of cases (1 = hemidemisemiquaver triplet or 64th note triplet). So 3/64 is an unusually short duration, which would be worth investigating. It may be an encoding error or it may be an ornament that could be removed.