[Hpr] tags, virtualization, virtualisation

Dave Morriss perloid at autistici.org
Tue Aug 21 08:54:26 PDT 2018

On 19/08/18 04:56, Mike Ray wrote:
> How about using the function of espeak to get the phonemes for a tag?
> The result in calling espeak for virtuallisation (the right way to spell
> it) and virtualization (nasty, left-pondian way of spelling it ;- ) is
> surely the same.
> I once did this in a commercial database to provide a kind of fuzzy
> search.  Works well.
> I did it by adding a column to each table where searches done on names
> was done and populating that column with the phonemes returned by
> espeak, after removing 'noise' words like 'the' etc.
> So, if there is a one-to-many relationship between show and tags, IOW
> show number XXX has one row in a table but as many rows as there are
> tags for it in the tags table, the tags table would have an extra
> column, one for the tag and one for it's phoneme.  Then search on the
> phoneme, not the literal tag.

Yes. Of course bjb has a point which I have largely ignored up until now ;-)

I like Mike's idea. I built an LDAP system for the university I used to
work for, where it was used for looking up people by name - amongst
other functions. I included a soundex encoded version of names in one
index, which could be searched for through the web front-end.

I don't know how soundex compares to what espeak can generate, but it
would be interesting to look at these issue.


Dave Morriss, Edinburgh, Scotland, UK | perloid at autistici.org

More information about the Hpr mailing list