fix: librispeech waveform dtype

archinetai · Sep 14, 2022 · 1147299 · 1147299
1 parent b2f3eaf
commit 1147299
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 2 deletions.
diff --git a/audio_data_pytorch/datasets/libri_speech_dataset.py b/audio_data_pytorch/datasets/libri_speech_dataset.py
@@ -30,7 +30,7 @@ def __getitem__(
     ) -> Union[Tensor, Tuple[Tensor, Dict]]:
         idx = idx.tolist() if torch.is_tensor(idx) else idx  # type: ignore
         data = self.dataset[idx]
-        waveform = torch.tensor(data["audio"]["array"]).view(1, -1)
+        waveform = torch.tensor(data["audio"]["array"]).view(1, -1).float()
         info = dict(
             sample_rate=data["audio"]["sampling_rate"],
             text=data["text"],

diff --git a/setup.py b/setup.py
@@ -3,7 +3,7 @@
 setup(
     name="audio-data-pytorch",
     packages=find_packages(exclude=[]),
-    version="0.0.11",
+    version="0.0.12",
     license="MIT",
     description="Audio Data - PyTorch",
     long_description_content_type="text/markdown",