Fix type for lm parameter in decoder (pytorch#2273)

Summary: Pull Request resolved: pytorch#2273 Reviewed By: mthrok Differential Revision: D34799335 Pulled By: carolineechen fbshipit-source-id: d0eea79448efdbd84758a3f433ab9350b4c94e91
xiaohui-zhang · May 4, 2022 · 688f752 · 688f752
1 parent a8906e7
commit 688f752
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/torchaudio/prototype/ctc_decoder/ctc_decoder.py b/torchaudio/prototype/ctc_decoder/ctc_decoder.py
@@ -181,7 +181,7 @@ def idxs_to_tokens(self, idxs: torch.LongTensor) -> List:
 def lexicon_decoder(
     lexicon: str,
     tokens: Union[str, List[str]],
-    lm: str = None,
+    lm: Optional[str] = None,
     nbest: int = 1,
     beam_size: int = 50,
     beam_size_token: Optional[int] = None,