The site has a simple API for automated queries: https://safe-reaches-13562.herokuapp.com/?f1=XXX&f2=XXX&f3=XXX
The values of F1 and F2 are obligatory; the two-vowel classifier will be used if no value for F3 is provided. The program does not allow for batch queries (all but the first values for each formant will be disregarded). Consider setting up a local copy if you have a large data-set to process.
The response is a JSON-formatted dictionary of the following form:
{
"predicted_vowel": "a",
"lang": {
"2295": "Neapolitan",
"1150": "Finnish",
"45": "Palestinian Arabic",
"1410": "Hadza",
"2829": "Russian",
"381": "Chickasaw",
"1258": "French"
},
"vowel": {
"2295": "a",
"1150": "\u00e6",
"45": "a",
"1410": "a",
"2829": "a",
"381": "a",
"1258": "a"
},
"f1": {
"2295": 604.0,
"1150": 578.0,
"45": 585.0,
"1410": 620.0,
"2829": 553.0,
"381": 587.0,
"1258": 615.0
},
"f2": {
"2295": 1513.0,
"1150": 1485.0,
"45": 1495.0,
"1410": 1485.0,
"2829": 1504.0,
"381": 1500.0,
"1258": 1465.0
},
"f3": {
"2295": 2458.452755905512,
"1150": 2535.0,
"45": 2458.452755905512,
"1410": 2458.452755905512,
"2829": 2514.0,
"381": 2452.0,
"1258": 2458.452755905512
}
}