feat: skip special tokens

descriptinc · Mar 6, 2024 · 11f77d0 · 11f77d0
1 parent f611479
commit 11f77d0
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/audiotools/core/whisper.py b/audiotools/core/whisper.py
@@ -72,7 +72,7 @@ def get_whisper_transcript(self, **kwargs) -> str:
             input_features = input_features.to(self.whisper_device)
             generated_ids = self.whisper_model.generate(input_features=input_features, **kwargs)
 
-        transcription = self.whisper_processor.batch_decode(generated_ids)
+        transcription = self.whisper_processor.batch_decode(generated_ids, skip_special_tokens=True)
         return transcription[0]
 
     def get_whisper_embeddings(self, **kwargs) -> torch.Tensor: