No public description

PiperOrigin-RevId: 702919949
google-research · Dec 5, 2024 · 720f3d2 · 720f3d2
1 parent 0567dc9
commit 720f3d2
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 0 deletions.
diff --git a/nisaba/scripts/natural_translit/brahmic/grapheme_inventory.py b/nisaba/scripts/natural_translit/brahmic/grapheme_inventory.py
@@ -131,6 +131,7 @@
 
 SYMBOL = [c.Char(*args) for args in [
     ['ind', '.', ph.SIL],
+    ['sgn', '-', ph.SIL],
     ['zwj', '+', ph.SIL],
     ['zwn', '|', ph.SIL],
 ]]

diff --git a/nisaba/scripts/natural_translit/language_params/mr.py b/nisaba/scripts/natural_translit/language_params/mr.py
@@ -46,6 +46,8 @@
 
 
 _TXN_OPS = fl.FstList(
+    g2p.AI_TO_EH_LONG,
+    g2p.AU_TO_OH_LONG,
     g2p.A_TO_EC,
     g2p.VOCALIC_U,
     g2p.H_ASP,

diff --git a/nisaba/scripts/natural_translit/romanization/testdata/mr_iso_nat.textproto b/nisaba/scripts/natural_translit/romanization/testdata/mr_iso_nat.textproto
@@ -65,3 +65,18 @@ rewrite {
   input: "māṛʰā"
   output: "madha"
 }
+rewrite {
+  rule: "ISO_TO_NAT"
+  input: "aisā"
+  output: "aisa"
+}
+rewrite {
+  rule: "ISO_TO_NAT"
+  input: "kannauja"
+  output: "kannauj"
+}
+rewrite {
+  rule: "ISO_TO_NAT"
+  input: "a-ēka"
+  output: "aek"
+}