Overview

This library may be used to convert numerals to decimals; it is intended to be used to process the dictation produced by a speech-recognition engine such as the one in Dragon.

Public API

The API consists of just one function 'toDecimal' which takes a single string argument.

Chunking

The purpose of chunking is to make numbers easier to enter. For example, saying 'ten ten ten' is easier than saying 'one hundred one thousand ten' to enter the number 101010. In the first case, the chunk size is 2, but in the second case, the chunk size is 6.

>>> num_dec.toDecimal("ten ten ten")
'101010'

>>> num_dec.toDecimal("one hundred one thousand ten")
'101010'

Converting decimals to numerals by chunking is not always invertible; 970053->9700,53->ninety seven hundred fifty three->9753!=970053. This issue can be addressed by simply pronouncing zeros; 970053->97,0,0,53->ninety seven zero zero fifty three->970053.

>>> num_dec.toDecimal("ninety seven hundred fifty three")
'9753'

>>> num_dec.toDecimal("ninety seven zero zero fifty three")
'970053'

Usage Examples

The function 'toDecimal'

1. supports chunking:

   >>> num_dec.toDecimal("four five nine two")
   '4592'

2. does not alter non-numerals:

   >>> num_dec.toDecimal("I'll have thirteen eggs for breakfast.")
   "I'll have 13 eggs for breakfast."

3. can handle long dictations:

   >>> num_dec.toDecimal("one two three four five six seven eight nine ten")
   '12345678910'

Installation

On most Linux machines, the following command may be used:

$ sudo install -D num_dec.py /usr/share/python/num_dec/num_dec.py

Testing

The API methods were tested for the first 1 billion integers with the chunk size ranging between -4 and 1, where a negative chunk size means that digits were grouped from right to left. The tests may be rerun for all natural numbers from 0 to 10000 with the following command:

>>> num_dec.testPackage(10000)

Removal

On most Linux machines, the following may be used to remove num_dec:

$ sudo rm -rf /usr/share/python/num_dec