thumb|PHOIBLE logo PHOIBLE (short for "Phonetics Information Base and Lexicon") is a linguistic database accessible through its website and compiling phonological inventories from primary documents and tertiary databases into a single, easily searchable sample. The 2019 version 2.0 includes 3,020 inventories containing 3,183 segment types found in 2,186 distinct languages. It is edited by Steven Moran, Assistant Professor from the Institute of Biology at the University of Neuchâtel and Daniel McCloy, Researcher at the Institute for Learning and Brain Sciences at the University of Washington.
thumb|PHOIBLE logo PHOIBLE (short for "Phonetics Information Base and Lexicon") is a linguistic database accessible through its website and compiling phonological inventories from primary documents and tertiary databases into a single, easily searchable sample. The 2019 version 2.0 includes 3,020 inventories containing 3,183 segment types found in 2,186 distinct languages. It is edited by Steven Moran, Assistant Professor from the Institute of Biology at the University of Neuchâtel and Daniel McCloy, Researcher at the Institute for Learning and Brain Sciences at the University of Washington.
== Principles of PHOIBLE == Rather than imposing a single system of describing languages, PHOIBLE attempts to be faithful to the various description methods found in source documents (often called "doculects") and to encode all character data in a consistent representation according to the Unicode API.
Discovered by embedding cosine similarity (sentence-transformers MiniLM, 384-dim).