On Tue, Aug 21, 2018, 10:56 AM Dave Morriss <[email protected]> wrote:
> I don't know how soundex compares to what espeak can generate, but it > would be interesting to look at these issue. > Soundex is fairly simplistic. It's a four character code that captures the first letter of a word and the next three consonants, converting them into numbers based on Anglocentric phonetic groupings. Double metaphone is a similar algorithm that uses a little more context and supports a few more pronunciations common to other languages. I wonder if a more flexible approach might be to support a notion of tag "synonyms". I haven't thought through an elegant way to store them in the DB, but the rough idea is that a show's tags are stored as-is, but when displaying or searching, its "effective tags" are the union of its "stored tags" and any synonyms for those tags. Of choose this comes at the cost of manually managing those synonyms--unless every new tag automatically gets its soundex (or soundex and espeak phoneme list and double metaphone) as synonym. >
_______________________________________________ Hpr mailing list [email protected] http://hackerpublicradio.org/mailman/listinfo/hpr_hackerpublicradio.org
