Get the latest tech news
Phonetic Matching
Recently, I’ve spent quite a lot of time working on Storyteller. It’s an open source, self-hostable platform for automatically syncing audiobooks and ebooks. You give Storyteller the ebook and audiobook files for the same book, and it spits out a new ebook file with the audio embedded, such that the text can be highlighted while the audio is playing. In order to do this, Storyteller has to answer a pretty challenging question: How do you automatically align the text with the corresponding audio?
In 1990, Lawrence Phillips — an analyst working at an insurance agency at the time — developed an alternative to Soundex, called Metaphone, which attempts to index words (not just names) by their English pronunciation. In 2008, mathematician and genealogist Alexander Beider and computer and software engineer Stephen Morse developed the Beider-Morse Phonetic Matching algorithm to aid in their Jewish genealogical research. Then, when the Ellis Island database was put online, he used the Datch-Mokotoff Soundex algorithm to search for names that sounded like “Matinsky”, his grandfather’s surname before he got married and changed it to Morse.
Or read this on Hacker News