Skip to content

Text-to-speech

Abbreviations, unusual pronunciations, large digits and ordinals

Abbreviations, unusual pronunciation and large digits are common to GTFS text fields. In the example below for TriMEt, we can see how the text-to-speech field should be used:

  • Abbreviations are fully spelled out: e.g. “SW” becomes “southwest”; “Ave” becomess “avenue”.
  • Pronunciations are spelled in a way that the software reads them correctly: e.g. “Orenco” becomes “orrainkoe”; “Merlo” becomes “murlo”.
  • Large digits are spelled out as they would be said: “3300” becomes “thirty-three hundred”. Otherwise, the software would read “3300” as “three thousand three hundred”.
  • Ordinals, such as 1st, 2nd and 3rd, should be spelled out: e.g. “1st” becomes “first”.

stops.txt

stop_id stop_name tts_stop_name
9163 SW 125th & Longhorn southwest one hundred twenty fifth & longhorn
9836 Orenco MAX Station orrainkoe max station
9828 Merlo Rd/SW 158th Ave MAX Station murlo road southwest one hundred fifty eighth avenue max station
10074 3300 Block NW 35th thirty-three-hundred block northwest thirty fifth

Acronyms

For acronyms that are referred to by their letters, letters should be followed by periods or separated by spaces. This clarifies that the acronym should be read letter-by-letter, and not as a word.

For Tampa, the headsign “North to UATC” contains an acronym that is pronounced by its individual letters. The text-to-speech disambiguation would be:

trips.txt

trip_headsign tts_trip_headsign
North to UATC north to u.a.t.c.

or

trip_headsign tts_trip_headsign
North to UATC north to u a t c

Conversely, some acronyms should be read as words: e.g. NATO; NASA. The text-to-speech field should reflect this.

Note

The field trips.tts_trip_headsign is not yet official in the specification.

Clarifying abbreviations with multiple meanings

The “St” abbreviation has multiple meanings: “street”, “saint”, “station” and “1st” to mean “first”. The text-to-speech field can address these double meanings by spelling out the correct word, and doing so in a way that is legible by the TTS software.