Current state : LinguaLibre's naming convention is largely based on - as the main field separator.
( and ) are also field informers.

Review

LinguaLibre Queries Services let you see the current situation.

List of speaker by name and presense of - as separator :

sparql
SELECT *
WHERE {
  ?id prop:P2 entity:Q3 .
  SERVICE wikibase:label {
    bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en" .
    ?id rdfs:label ?name .
  }
  BIND (regex(STR(?name),"-") AS ?has_separator)
}
ORDER BY DESC (?has_separator)

How many:

sparql
SELECT ?has_separator (COUNT(?has_separator) AS ?found)
WHERE {
  ?id prop:P2 entity:Q3 .
  SERVICE wikibase:label {
    bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en" .
    ?id rdfs:label ?name .
  }
  
  BIND (regex(STR(?name),"-") AS ?has_separator)
  # filter( regex(?name, "-" ))
}
#ORDER BY DESC (?has_separator)
GROUP BY (?has_separator)

51/1000 (5%) of the speakers' username contain -, which makes regex on their files more unpredictable. A better field separator would be welcome.

Suggestion

On peut/doit prendre un plus rare, a minima qui n'est pas un des caractères présents sur nos claviers.
U+FF0D － FULLWIDTH HYPHEN-MINUS LL－Q150 (fra)－Roll-Morton－vert.wav
Ou doubler le séparateur, pour créer une pratique unique. LL--Q150 (fra)--Roll-Morton--vert.wav