Merge pull request #29 from opensource-spraakherkenning-nl/add-bn-nl-…

…variations

Add variations from top 20 confusion pairs

ASR_NL_benchmark/interface.py

@@ -22,8 +22,6 @@ def upload_page():
         kind = request.form.get('kind')
         skip_ref_norm = request.form.get('skip-ref-norm')
         skip_hyp_norm = request.form.get('skip-hyp-norm')
-        print(skip_hyp_norm)
-        print(skip_ref_norm)
         global benchmarking
         benchmarking = pipeline.Pipeline(hyp, 'ctm', ref, 'stm', kind, skip_ref_norm, skip_hyp_norm)
         Thread(target=benchmarking.main).start()

ASR_NL_benchmark/variations.glm

@@ -13,8 +13,6 @@ z'n => zijn / [ ] __ [ ]
 'k => ik / [ ] __ [ ]
 'r => er / [ ] __ [ ]
 'ns => eens / [ ] __ [ ]
-ie => hij / [ ] __ [ ]
-da's => dat is / [ ] __ [ ]
 d'ruit => eruit / [ ] __ [ ]
 restaurant- => restaurant / [ ] __ [ ]
 jeugd- => jeugd / [ ] __ [ ]
@@ -59,7 +57,6 @@ tewerk => te werk / [ ] __ [ ]
 [concept-] => [{ concept- / concept }] / [ ] __ [ ]
 [NAVO-] => [{ NAVO- / NAVO }] / [ ] __ [ ]
 [uh] => [{ uh / %HESITATION }] / [ ] __ [ ]
-[BNR-nieuwsradio] => [{ BNR-nieuwsradio / BNR nieuwsradio }] / [ ] __ [ ]
 ;;
 ;; BN-VL
 [Darfour] => [{ Darfour / Darfur }] / [ ] __ [ ]
@@ -90,3 +87,14 @@ tewerk => te werk / [ ] __ [ ]
 [marktonderzoekbureau] => [{ marktonderzoekbureau / marktonderzoeksbureau }] / [ ] __ [ ]
 [Noordwestkust] => [{ Noordwestkust / Noord-Westkust }] / [ ] __ [ ]
 [carnavalvierders] => [{ carnavalvierders / carnavalsvierders }] / [ ] __ [ ]
+
+;; Whisper evaluation on N-Best
+;; BN-NL
+ie => hij / [ ] __ [ ]
+da's => dat is / [ ] __ [ ]
+[BNR-nieuwsradio] => [{ BNR-nieuwsradio / BNR nieuwsradio }] / [ ] __ [ ]
+[Moszkowicz] => [{ Moszkowicz / Moskovic }] / [ ] __ [ ]
+[Kooi] => [{ Kooi / Kooij }] / [ ] __ [ ]
+[Araújo] => [{ Araújo / Araujo }] / [ ] __ [ ]
+[Bagdad] => [{ Bagdad / Baghdad }] / [ ] __ [ ]
+[Holleeder] => [{ Holleeder / Holleder }] / [ ] __ [ ]