add: mean and std

musdb18.py

@@ -36,50 +36,73 @@ class Musdb18Dataset(datasets.GeneratorBasedBuilder):
     def _info(self):
         return datasets.DatasetInfo(
             description=_DESCRIPTION,
-            features=datasets.Features({
-                "name": datasets.Value("string"),
-                "n_window": datasets.Value("int16"),
-                **{name: datasets.Audio(sampling_rate=self.SAMPLING_RATE, mono=False)
-                    for name in self.INSTRUMENT_NAMES}
-            })
+            features=datasets.Features(
+                {
+                    "name": datasets.Value("string"),
+                    "n_window": datasets.Value("int16"),
+                    **{
+                        name: datasets.Audio(
+                            sampling_rate=self.SAMPLING_RATE, mono=False
+                        )
+                        for name in self.INSTRUMENT_NAMES
+                    },
+                    "mean": datasets.Value("float"),
+                    "std": datasets.Value("float"),
+                }
+            ),
         )
 
     def _split_generators(self, dl_manager):
         #! you must have your folder locally!
         archive_path = dl_manager.download_and_extract(
-            "https://zenodo.org/record/1117372/files/musdb18.zip?download=1")
+            "https://zenodo.org/record/1117372/files/musdb18.zip?download=1"
+        )
 
         return [
             datasets.SplitGenerator(
                 name=datasets.Split.TRAIN,
-                gen_kwargs={
-                    "audio_path": f"{archive_path}/train"}
+                gen_kwargs={"audio_path": f"{archive_path}/train"},
             ),
             datasets.SplitGenerator(
                 name=datasets.Split.TEST,
-                gen_kwargs={
-                    "audio_path": f"{archive_path}/test"
-                }
-            )
+                gen_kwargs={"audio_path": f"{archive_path}/test"},
+            ),
         ]
 
     def _generate_stem_dict(self, S, song_name, end):
-        return {name: {"path": f"{song_name}/{name}",
-                       "array": S[i, end-self.WINDOW_SIZE:end, :], "sampling_rate": self.SAMPLING_RATE}
-                for i, name in enumerate(self.INSTRUMENT_NAMES)}
+        return {
+            name: {
+                "path": f"{song_name}/{name}",
+                "array": S[i, end - self.WINDOW_SIZE : end, :],
+                "sampling_rate": self.SAMPLING_RATE,
+            }
+            for i, name in enumerate(self.INSTRUMENT_NAMES)
+        }
 
     def _generate_examples(self, audio_path):
         id_ = 0
         for stems_path in Path(audio_path).iterdir():
             song_name = stems_path.stem
             S, sr = stempeg.read_stems(
-                str(stems_path), dtype=np.float32, multiprocess=False)
+                str(stems_path), dtype=np.float32, multiprocess=False
+            )
+
+            mixture = S.sum(axis=0).T
+            assert mixture.shape[0] == 2
+            # from (n_instr, n_chann, n_samp) -> (n_chann, n_samp)
+            mixture = mixture.mean(0)  # channel_wise mean -> (n_samples,)
+            mean = mixture.mean().item()
+            std = mixture.std().item()
 
-            for idx, end in enumerate(range(self.WINDOW_SIZE, S.shape[1], self.WINDOW_SIZE)):
+            for idx, end in enumerate(
+                range(self.WINDOW_SIZE, S.shape[1], self.WINDOW_SIZE)
+            ):
                 yield id_, {
                     "name": song_name,
                     "n_window": idx,
-                    **self._generate_stem_dict(S, song_name, end)
+                    **self._generate_stem_dict(S, song_name, end),
+                    "mean": mean,
+                    "std": std,
                 }
 
                 id_ += 1
@@ -87,8 +110,10 @@ class Musdb18Dataset(datasets.GeneratorBasedBuilder):
             # It's very rare for song to have exactly 3 minutes
             yield id_, {
                 "name": song_name,
-                "n_window": idx+1,
-                **self._generate_stem_dict(S, song_name, end=S.shape[1])
+                "n_window": idx + 1,
+                **self._generate_stem_dict(S, song_name, end=S.shape[1]),
+                "mean": mean,
+                "std": std,
             }
 
             id_ += 1