Differences between old and current FAVE-extract outputs (error accessing newly pulled formant files)

[thealk]

Just updated bigbro9uk to include newest formant files... when running plotmish encountered following indexError (this is running on an older version of plotmish that has been working fine all day):

McGills-MacBook-Pro:plotmish mcgillLing$ python plotmish.py /Users/mcgillLing/3_MLML/Big\ Brother/bigbro9uk/data/formants_full /Users/mcgillLing/3_MLML/Big\ Brother/wavs -k Lisa
Found both text files and plotnik files
Which file type do you want to use?
(p)lotnik or (t)ext: t
Traceback (most recent call last):
File "plotmish.py", line 1234, in
main()
File "plotmish.py", line 676, in main
else: getVowelsTxt()
File "plotmish.py", line 367, in getVowelsTxt
moreForms = [re.sub('[[]]','',m).split(',') for m in v[36].split('],[')]
IndexError: list index out of range

[msonderegger]

Can you see what file is loading in plotmish when you get this?

[thealk]

Yep, it's the first ep file.

[thealk]

Just tried with another speaker to be sure; same problem - crashes at first episode.

[msonderegger]

is this the latest plotmish? not sure if this would matter, but you need the annotator in the call. This works for me:

python plotmish.py ~/bigbro/bigbro9uk/data/formants_full/ ~/bigbro/clipLinks/ ms -k ep90

[thealk]

no i hadn't pulled the newest plotmish - if it's ready to go i'll try that

[msonderegger]

yup, it works now.

[thealk]

same error but if it's working for you it must be something on my end. stay tuned.

[thealk]

keyword ep90 works for me... it seems to be certain episodes. ep80 also works, as well as ep8 (so episodes80-89), but ep0, ep02, ep03, ep10.... etc do not work. I haven't been able to figure out what it is about some of the files that's causing it to crash yet but if you have thoughts let me know. I'll continue with the dru and a subset of the eps for now...

[mishaschwartz]

They're all working for me. Which computer are you running it on? I can try to screen share in and see if there's something local that's screwing things up if you want.

[thealk]

I was running it on my own but i just tried running it with the ep clips on parmesan and same problem...

[thealk]

parmesan:plotmish mlml$ python plotmish.py /Users/mlml/bigbro9uk/data/formants_full /Users/mlml/bbFiles/clips/ep tk -k Lisa
Traceback (most recent call last):
File "plotmish.py", line 1258, in
main()
File "plotmish.py", line 700, in main
else: getVowelsTxt()
File "plotmish.py", line 391, in getVowelsTxt
moreForms = [re.sub('[[]]','',m).split(',') for m in v[36].split('],[')]
IndexError: list index out of range

[thealk]

(dru works fine)

[mishaschwartz]

Ok the problem is that they updated FAVE-extract and I used the new version to extract some of the ep files. The output for new FAVE-extract is slightly different so the columns don't line up. I'm going to make a quick fix for now and then I'll make a more permanent fix later. I'll comment and close this when the quick fix is going so that you can carry on with annotation.

[mishaschwartz]

Never mind, turns out the new fave-extract leaves out some crucial things like preceding and following segment as well. I think it'll actually be faster if we re-extract the formants with the old fave-align for our purposes now and we can figure out how we want to deal with this long term later.

[msonderegger]

Good catch -- that's annoying of them to change the output format. What you suggest sounds good -- could you do the re-extraction with old fave-extract?

(Less important: any idea why I'm not getting the same error as Thea on my machine when I do the same call?)

[msonderegger]

could one of you create github issues and/or basecamp todo's for figuring out how to deal with this longer-term?

[mishaschwartz]

The remeasured ep formant files are up in formants_full now.
Morgan, I have no idea why it was working for you and not for Thea. My only guess is that you had a local version that was actually extracted with the old fave-extract? Were you reading the formant.txt files from formants_full when you were using plotmish?

[msonderegger]

Hm, weird. I don't think I had a local version of the ep formant files
extracted with the old fave-extract... I don't think I've actually used
fave-extract myself for a while (rather than using the formant files you
upload). Anyway, as long as it works now, it's fine.

So just to be clear, formant files for all dru and ep clips are now in
formants_dru and formants_ep? If so, could you go ahead and delete
formants/ and formants_full/, to clean things up?

Morgan

On Wed, May 28, 2014 at 9:25 AM, mishaschwartz [email protected]:

The remeasured ep formant files are up in formants_full now.
Morgan, I have no idea why it was working for you and not for Thea. My
only guess is that you had a local version that was actually extracted with
the old fave-extract? Were you reading the formant.txt files from
formants_full when you were using plotmish?

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/15#issuecomment-44430967
.

[mishaschwartz]

No. They're all in formants_full. I'll put them in formants ep and dru and delete formants and formants full.

[mishaschwartz]

Done

[thealk]

Why did we want to get rid of formants_full? can I re-add it? Or is this a case where you'd rather use symbolic links?

[thealk]

also, in other news: ep files work for me now so yay!

[msonderegger]

Sure, go ahead and re-add it... this was my fault, I caused some confusion.

Morgan

On Thu, May 29, 2014 at 3:11 PM, thealk [email protected] wrote:

also, in other news: ep files work for me now so yay!

—
Reply to this email directly or view it on GitHub
#15 (comment).

[thealk]

added a todo on basecamp about the fave-extract issue and edited the title to this issue, so we can keep it open here.