cahya
/

wav2vec2-base-turkish

Automatic Speech Recognition

generated_from_trainer

hf-asr-leaderboard

robust-speech-event

Inference Endpoints

Model card Files Files and versions Community

cahya commited on Feb 3, 2022

Commit

2cf743a

•

1 Parent(s): 15a59f8

remove unigrams

Files changed (1) hide show

run_evaluation.py +1 -3

run_evaluation.py CHANGED Viewed

@@ -22,9 +22,7 @@ class KenLM:
         vocab_dict = tokenizer.get_vocab()
         self.vocabulary = [x[0] for x in sorted(vocab_dict.items(), key=lambda x: x[1], reverse=False)]
         self.vocabulary = self.vocabulary[:-1]
-        with open(unigrams, "r") as f:
-            unigrams_text = f.read()
-            self.decoder = build_ctcdecoder(self.vocabulary, model_name, unigrams=unigrams_text)
     @staticmethod
     def lm_postprocess(text):

         vocab_dict = tokenizer.get_vocab()
         self.vocabulary = [x[0] for x in sorted(vocab_dict.items(), key=lambda x: x[1], reverse=False)]
         self.vocabulary = self.vocabulary[:-1]
+        self.decoder = build_ctcdecoder(self.vocabulary, model_name)
     @staticmethod
     def lm_postprocess(text):