pydrobert.speech.config

Package constants used throughout pydrobert.speech

pydrobert.speech.config.EFFECTIVE_SUPPORT_THRESHOLD = 0.0005

Value considered roughly zero for support computations

No function is compactly supported in both the time and Fourier domains, but large regions of either domain can be very close to zero. This value serves as a threshold for zero. The higher it is, the more accurate computations will be, but the longer they will take

Type:

float

pydrobert.speech.config.LOG_FLOOR_VALUE = 1e-05

Value used as floor when taking log in computations

Type:

float

pydrobert.speech.config.SOUNDFILE_SUPPORTED_FILE_TYPES

A list of the types of files SoundFile will be responsible for reading

If soundfile can be imported, it’s the intersection of soundfile.available_formats() with the set “wav”, “ogg”, “flac”, and “aiff”.

See also

pydrobert.speech.util.read_signal

Where this flag is used

Type:

set

pydrobert.speech.config.USE_FFTPACK

Whether to use scipy.fftpack

The scipy implementation of the FFT can be much faster than the numpy one. This is set automatically to True if scipy.fftpack can be imported. It can be set to False to use the numpy implementation.

Type:

bool