The radio broadcast that has the most impact on our lives comes from NOAA Weather Radio. It is almost always the first broadcast we tune into each day, and we frequently consult it throughout every day. While primarily a source of information, it often becomes a source of entertainment as well. Most of this comes from the voices on the broadcast, which are not human, but computer-generated.
Computer-generated voices have delivered our weather forecasts for years. Since we’ve moved to the homestead, a system upgrade significantly improved the voice quality. Formerly, a male voice with a strange, syllable-swallowing accent dominated the broadcast. He was replaced by male and female voices that are almost indistinguishable from human. Oddly, the early voice appears to deliver the current time just before the marine forecast.
The new voices are so good that when they make mistakes, it’s rather comical. Usually, this comes from too-literal readings of the typed forecast. Individual slashes in the text are commonly broadcast as the word “slash.” The Canadian forecast, which is used to help cover the passes to the north, have some sort of discrepancy in the word “sun” that makes the computer read it as an abbreviation for Sunday. This leads to predictions of “mixed Sunday and cloud,” leading us to joke that the wind will be so strong it’ll blow us into next week.
Place names cause a lot of strange mistakes, which is odd, because NOAA seems to have the ability to “teach” the voices correct pronunciation. The new voices never seem to have trouble with Haida Gwaii, although the same can still not be said for some of our local, live announcers. And yet, Kenai (“KEEN-eye”) gets pronounced “Ken-EYE” by the computer, so it’s a mixed bag.
Lately, the computer has begun confusing the word “wind,” as in air movement, with the word “wind,” as in “wind a watch.” I believe the problem comes from a change in wording in the forecasts. Where they used to say “South winds to X knots,” they now say “South winds up to X knots.” The voice program apparently finds it statistically more probable that the word “wind” followed immediately by “up” will be pronounced with a long i rather than a short i.
I don’t know how long it might take for them to catch and correct this. In the meantime, it’s a bit of entertainment to go along with our information.