One of the great thing about the Penn Forced Aligner is that we can add to its dictionary (…p2fa/model/dict) (“dict” is the file, it has no extension and is plaintext).
A custom dictionary is a list of entries like:
WORD T R A N S C R I P T I O N
ANOTHER B L A H
Information for creating custom dictionaries
- CAPITALIZE EVERYTHING
- There are two spaces between a word and its transcription
- Each phone is separated by a space
- Only use the CMU codes in the attached file
- Be careful to separate consonants, regardless of their status in the target language
- All vowels are of the form XXn, where n represents stress; I use 0 (zero) as default because it’s not important
- [h]’s code is HH, it is the only ‘trick’ consonant
- Always include the special CMU codes sp, {br}, {cg}, etc., they are at the end of the dictionary file and should be copied into your new dictionary
Currently we have
- An English dictionary modified with extra words
- A German dictionary (GR)
- A Norwegian dictionary (LA)