Unknown Phone issue

36 views
Skip to first unread message

Sara Carta

unread,
Dec 13, 2021, 2:22:07 PM12/13/21
to FAVE (Force Alignment and Vowel Extraction) Users Group
Hello everyone, 

I am new to FAVE and I am now learning to use its recent adaptation for Danish and Swedish (Young, N. & McGarrah, M. (2021). Forced alignment for Nordic languages: Rapidly constructing a high-quality prototype. Nordic Journal of Linguistics). 

I can succesfully input the transcription and obtain an unknown.txt file, where words unknown to the dictionary are stored. 
At this point, as described in the FAVE tutorial (https://github.com/JoFrhwld/FAVE/wiki/Using-FAVE-align), I open the unknown.txt file using Excel and I insert the correct Arpabet (or, in this case, SweFAbet) equivalents for each unknown word, in the second column. Then I delete all columns, except for the first and second, I save it as a tab-delimited txt file (under the name of input.txt) and I run the code again in dictionary-check-mode. 
The problem is that I get a persistent "unknown phone error", which seems to be tied to the position of the word in the input.txt file and not to a specific actual mistake: whatever the word-order, it is always the second word that shows a phone error. Whenever I try to correct it on the fly, the correction does not fit and error messages continue to pop up. Even correcting directly on the file doesn't seem to work. 
And, in the end, no correction at all should be needed, since all the phones seem to be correct and already used in the dictionary. 
I am sending a screenshot of the situation attached to this email.

I would appreciate any insights or suggestions on this!
Thank you for your time. 

Best regards, 

Sara 

Error.JPG

beke_...@hotmail.com

unread,
Dec 14, 2021, 12:29:18 PM12/14/21
to FAVE (Force Alignment and Vowel Extraction) Users Group

Hi Sara,
I've just read your post and checked out the error message.
I'm not familiar with the adaptation of FAVE called "SweFA" but I've scanned the article you referred to in your post. To me it seems as if the combination "UOH0" is not possible given the definition of sounds on p. 24 of the article. I think what would be possible is "UH0".  I'd also think that "OEEH1" has to be changed to "OEH1". 

I'm not sure if that helps. Maybe other people who know more about SweFA can confirm or add.

Best wishes
Beke

Reply all
Reply to author
Forward
0 new messages