[OSM-talk] handling street names in speech

Maarten Deen mdeen at xs4all.nl
Wed Jul 17 07:56:57 UTC 2019


On Wed, Jul 17, 2019 at 12:58 AM Andrew Errington <erringtona at gmail.com> 
wrote:

> I think this is a rendering issue (i.e. rendering speech instead of
> graphics) and as such does not belong in OSM.
> 
> The work to convert an arbitrary string into speech belongs in the
> TTS engine.
> 
> If we start putting IPA strings in OSM then we will start getting
> arguments about the "correct" pronunciation. At the very least it is
> tagging for the renderer, which we should avoid.

And if not, then you're at the will of the TTS engine. There are words 
that are pronounced differently depending on their meaning. Some example 
from the English language, I'm sure other languages have examples too:
a bow - to bow (second sounds more like baw)
he does things - the does do things (short o, long o)
A minute part of a minute
wind (is blowing) - wind (the clock)

And then we haven't even touched the problem on which syllable to put 
emphasis.

All these things make it impossible for a TTS engine to know what 
pronunciation is correct, except when you lay it down for each word. We 
had a new operator for the public transport in my neighborhood and they 
used a TTS engine to generate the stop information. It was just 
horrible. Incorrect pronunciation, incorrect emphasis.

The OsmAND TTS also does funny things, there is a John F. Kennedystreet 
in my town. The TTS pauses at the dot. Obviously it thinks the sentence 
ends there. But it doesn't. It's just an abbreviation. And no, this is 
not an abbreviation like St. or Ave. that you would write out fully. The 
street name is not John Fitzgerald Kennedystreet.

IMHO an IPA string would be a welcome and usefull addition to OSM.
Could you also not use it for transliterations?

Regards,
Maarten



More information about the talk mailing list