CMUdict is a versioned python wrapper package for The CMU Pronouncing Dictionary data files. The main purpose is to expose the data with little or no assumption on how it is to be used.

cmudict is available on PyPI. Simply install it with pip:

pip install cmudict

The cmudict data set includes 4 data files: cmudict.dict, cmudict.phones, cmudict.symbols, and cmudict.vp. See The CMU Pronouncing Dictionary for details on the data. Chances are, if you're here, you already know what's in the files.

Each file can be accessed through three functions, one which returns the raw (string) contents, one which returns a binary stream of the file, and one which does minimal processing of the file into an appropriate structure:

>>> import cmudict

>>> cmudict.dict() # Compatible with NLTK
>>> cmudict.dict_string()
>>> cmudict.dict_stream()

>>> cmudict.phones()
>>> cmudict.phones_string()
>>> cmudict.phones_stream()

>>> cmudict.symbols()
>>> cmudict.symbols_string()
>>> cmudict.symbols_stream()

>>> cmudict.vp()
>>> cmudict.vp_string()
>>> cmudict.vp_stream()

Three additional functions are included to maintain compatibility with NLTK: cmudict.entries(), cmudict.raw(), and cmudict.words(). See the nltk.corpus.reader.cmudict documentation for details:

>>> cmudict.entries() # Compatible with NLTK
>>> cmudict.raw() # Compatible with NLTK
>>> cmudict.words() # Compatible with NTLK

And finally, the license for the cmudict data set is available as well:

>>> cmudict.license_string() # Returns the cmudict license as a string

Built on or modeled after the following open source projects:

• The CMU Pronouncing Dictionary
• NLTK