Match Type — Sounds Alike

Match Type — Sounds Alike

Sounds Alike is a phonetic matching method that identifies values which sound similar when spoken aloud, even if they are spelled quite differently. This is particularly effective for matching personal names and place names.

How Sounds Alike Works

Sounds Alike uses phonetic algorithms such as Soundex and Metaphone to convert text values into phonetic codes. Two values that produce the same or similar phonetic codes are considered to sound alike, regardless of their spelling.

Examples of values that would match:

• Smith and Smyth — different spellings, same pronunciation
• Catherine and Katherine — the "C" and "K" produce the same sound
• Steven and Stephen — "v" and "ph" sound the same
• Meier and Meyer — common surname variant
• Thompson and Thomson — the silent "p" does not affect the sound

Configuring Minimum Sound Similarity

The Minimum Sound Similarity threshold controls how closely two values must sound alike to be considered a match. A higher threshold requires a closer phonetic match; a lower threshold allows more variation.

Best Use Cases

Sounds Alike is ideal for:

• First names — many names have multiple accepted spellings (Sara/Sarah, Jon/John, Anne/Ann)
• Last names — surname variants are extremely common (Schmidt/Smith, Garcia/Garsia)
• City names — phonetic matching catches misspellings of place names

Limitations

Phonetic matching has some limitations to keep in mind:

• It works best for English and Western European names. Names from other language families may not be encoded accurately.
• Very short names (2-3 characters) may produce ambiguous phonetic codes.
• It does not handle abbreviations or acronyms — "Robert" and "Bob" do not sound alike to the algorithm, even though they refer to the same name.

Combining with Fuzzy Matching

For the best name-matching results, consider using Sounds Alike for one criterion and Similar Text (https://help.matchlogic.io/article/258-match-type-similar-text) for another within the same definition. This combination catches both phonetic variants and typographical errors. For example:

• Criterion 1: First Name — Sounds Alike (catches "Catherine" / "Katherine")
• Criterion 2: Last Name — Similar Text at 80% (catches "Johnsn" / "Johnson")

Alternatively, create separate definitions — one using phonetic matching and another using fuzzy matching — so that records matching either approach are flagged as duplicates. See #creating-multiple-definitions for how to set this up.