Merge pull request #13 from opensource-spraakherkenning-nl/add-variat…

…ions Add variation + add back dash (-) to punctuation exceptions
opensource-spraakherkenning-nl · Nov 13, 2023 · fc17e2c · fc17e2c
2 parents 86ec141 + 655cf48
commit fc17e2c
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 1 deletion.
diff --git a/ASR_NL_benchmark/normalize.py b/ASR_NL_benchmark/normalize.py
@@ -44,7 +44,7 @@ def replace_numbers_and_symbols(text):
     >>> replace_numbers_and_symbols('12,3%')
     'twaalf komma drie procent'
     """
-    removed_punct = string.punctuation.replace("'", '')
+    removed_punct = string.punctuation.replace("'", '').replace('-', '')
     text_without_symbols = replace_symbols(text)
     clean_text = replace_numbers(text_without_symbols)
     clean_text = clean_text.translate(str.maketrans('', '', removed_punct))

diff --git a/ASR_NL_benchmark/variations.glm b/ASR_NL_benchmark/variations.glm
@@ -59,6 +59,7 @@ tewerk => te werk / [ ] __ [ ]
 [concept-] => [{ concept- / concept }] / [ ] __ [ ]
 [NAVO-] => [{ NAVO- / NAVO }] / [ ] __ [ ]
 [uh] => [{ uh / %HESITATION }] / [ ] __ [ ]
+[bnr-nieuwsradio] => [{ bnr-nieuwsradio / bnr nieuwsradio }]
 ;;
 ;; BN-VL
 [Darfour] => [{ Darfour / Darfur }] / [ ] __ [ ]