Fix encoding of CMUDict data reading (#1665)

ac1c9d5c · yangarbiter · GitHub · 43047865 · ac1c9d5c
Unverified Commit ac1c9d5c authored Aug 03, 2021 by yangarbiter Committed by GitHub Aug 03, 2021
Show whitespace changes
Inline Side-by-side

Showing with 1 addition and 1 deletion

torchaudio/datasets/cmudict.py torchaudio/datasets/cmudict.py +1 -1

No files found.
--- a/torchaudio/datasets/cmudict.py
+++ b/torchaudio/datasets/cmudict.py
@@ -148,7 +148,7 @@ class CMUDict(Dataset):
        with open(os.path.join(self._root_path, basename_symbols), "r") as text:
            self._symbols = [line.strip() for line in text.readlines()]

-        with open(os.path.join(self._root_path, basename), "r") as text:
+        with open(os.path.join(self._root_path, basename), "r", encoding='latin-1') as text:
            self._dictionary = _parse_dictionary(text.readlines(),
                                                 exclude_punctuations=self.exclude_punctuations)