- CASE vs. case definitely DOES matter
- use of ' or " has interesting (and unpredictable) results
- and testing these rigs isn't has hard as one might imagine
Of the following segment names:
- Code: Select all
1: “E” Rd, 2: 'E' Rd, 3: E. Rd, 4: “E.” Rd, 5: 'E.' Rd,
1: “e” Rd, 2: 'e' Rd, 3: e. Rd, 4: “e.” Rd, 5: 'e.' Rd, 6: e Rd,
Some were "gobbledegook", many were "East Rd" and only one actually delivered "Eee Road"
- Code: Select all
And while "Wash Rd", and "Wash. Rd" were pronounced as the word, "WASH Rd" somehow or another came out as "Washington Rd"
For "unknown" words, the TTS engine has two different behaviors:
- spell out the letters, e.g. TR = "tee ar"
- attempt pronounciation, e.g. "Tline" = "t'line" (or whatever you might imagine that to sound like)
I thought that this was varied based on UPPER CASE = spell it out, Mixed Case = attempt to pronounce, but this is not consistent.
As far as this being easier than anticipated to actually test, here are the tips:
- spreadsheets are your friend (this is generally always the case)
- Name the segments as I've done above, with multiple test values identified numerically (with colo) and comma separated
- I set up segment names with up to 10 different test values, and still only had to replay one or two more than once
- create the segments in a radial, coming out from one junction
- attach a small segment at the end of the "test named" segment with a short name. This makes it easy to select the destination.
- Most important, it isn't actually necessary to HEAR the tts on your phone! If you are close enough to the junction with your test segment, the .tts file will be created. "Stop nav" to that destination, navigate to the next, and so on. Not even necessary to move your car.... or even be in your car. I sat at the dining room table (just in front of my big bad junction)
This is about what mine looks like at 100m/500ft zoom. The northern "radial" was more than adequate for me to be able to select the destination segments