README

ColognePhonetic

A PHP OOP implementation of the "Kölner Phonetik Index".

This is an adaption of Andy Theiler's precedural implementation with some fixes and unit tests.

Requirements

• "PHP" >= "5.5"
• "ext-iconv" "*"

Installation

composer require "hollodotme/cologne-phonetic" "~1.0.0"

Basic usage

<?php

use hollodotme\ColognePhonetic\ColognePhonetic;

$inputCharset    = 'UTF-8';
$colognePhonetic = new ColognePhonetic( $inputCharset );

# Retrieve index for a single word
$wordIndex = $colognePhonetic->getWordIndex( 'Wort' );

var_dump( $wordIndex );

# Retrieve index for a phrase (multiple words)
# Phrase is split into words, returns an assoc. array with [ "word" => "index" ]
$phraseIndex = $colognePhonetic->getPhraseIndex( 'Ein Satz mit mehreren Wörtern' );

var_dump( $phraseIndex );

Output:

string(3) "072"

array(5) {
  'Ein' =>
  string(2) "06"
  'Satz' =>
  string(1) "8"
  'mit' =>
  string(2) "62"
  'mehreren' =>
  string(3) "676"
  'Wortern' =>
  string(5) "07276"
}

Algorithm

The "Kölner Phonetik" transliterates every character of a word to a numeric code between "0" und "8", considering at most one adjacent character as context. Some rules apply especially to the beginning of a word (initial sound). This way similar phonems get mapped to the same numeric code. For example the caracters "W" and "V" are both transliterated to "3". The phonetic code of "Wikipedia" is "3412". In contrast to the "Soundex-Code" the length of the "Kölner Phonetik Index" is not restricted.

The fact that "SC" has priority before "CH" is explained in the addition "but not after" on line 10 in the table above. This rule was not explicitly part of the official publication, but can be deduced implicitly from the published examples.

The transliteration happens in 3 steps:

1. Transliteration of characters from left to right like described in the table above.
2. Removing all duplicate codes.
3. Removing of all "0" codes except at the beginning.