diff --git a/lerobot/common/datasets/video_utils.py b/lerobot/common/datasets/video_utils.py
index d1b25023..c7a5da61 100644
--- a/lerobot/common/datasets/video_utils.py
+++ b/lerobot/common/datasets/video_utils.py
@@ -64,12 +64,12 @@ def decode_audio(
     if backend == "torchcodec":
         raise NotImplementedError("torchcodec is not yet supported for audio decoding")
     elif backend == "ffmpeg":
-        return decode_audio_torchvision(audio_path, timestamps, duration)
+        return decode_audio_torchaudio(audio_path, timestamps, duration)
     else:
         raise ValueError(f"Unsupported video backend: {backend}")
 
 
-def decode_audio_torchvision(
+def decode_audio_torchaudio(
     audio_path: Path | str,
     timestamps: list[float],
     duration: float,
diff --git a/tests/datasets/test_compute_stats.py b/tests/datasets/test_compute_stats.py
index 9cf9f760..2ebf95f2 100644
--- a/tests/datasets/test_compute_stats.py
+++ b/tests/datasets/test_compute_stats.py
@@ -87,7 +87,7 @@ def test_sample_audio_from_path(mock_load):
     assert len(audio_samples) == estimate_num_samples(16000)
 
 
-def test_sample_audio_from_data(mock_load):
+def test_sample_audio_from_data():
     audio_data = np.ones((16000, 2), dtype=np.float32)
     audio_samples = sample_audio_from_data(audio_data)
     assert isinstance(audio_samples, np.ndarray)