feat: stereo transform, convert always to 2 channels

README.md

@@ -117,8 +117,8 @@ resample = Resample(source=48000, target=22050), # Resamples from 48kHz to 22kHz
 from audio_data_pytorch import OverlapChannels
 overlap = OverlapChannels() # Overap channels by sum (C, N) -> (1, N)
 
-from audio_data_pytorch import DuplicateChannels
-duplicate = DuplicateChannels() # Duplicate channels (1, N) -> (2, N) or (2, N) -> (2, N)
+from audio_data_pytorch import Stereo
+stereo = Stereo() # Duplicate channels (1, N) -> (2, N) or (2, N) -> (2, N)
 
 from audio_data_pytorch import Scale
 scale = Scale(scale=0.8) # Scale waveform amplitude by 0.8

audio_data_pytorch/transforms/__init__.py

@@ -1,8 +1,8 @@
 from .all import AllTransform
 from .crop import Crop
-from .duplicate_channels import DuplicateChannels
 from .loudness import Loudness
 from .overlap_channels import OverlapChannels
 from .randomcrop import RandomCrop
 from .resample import Resample
 from .scale import Scale
+from .stereo import Stereo

audio_data_pytorch/transforms/all.py

@@ -4,12 +4,12 @@
 
 from ..utils import exists
 from .crop import Crop
-from .duplicate_channels import DuplicateChannels
 from .loudness import Loudness
 from .overlap_channels import OverlapChannels
 from .randomcrop import RandomCrop
 from .resample import Resample
 from .scale import Scale
+from .stereo import Stereo
 
 
 class AllTransform(nn.Module):
@@ -21,7 +21,7 @@ def __init__(
         random_crop_size: Optional[int] = None,
         loudness: Optional[int] = None,
         scale: Optional[float] = None,
-        duplicate_channels: bool = False,
+        use_stereo: bool = False,
         overlap_channels: bool = False,
     ):
         super().__init__()
@@ -39,7 +39,7 @@ def __init__(
             RandomCrop(random_crop_size) if exists(random_crop_size) else nn.Identity(),
             Crop(crop_size) if exists(crop_size) else nn.Identity(),
             OverlapChannels() if overlap_channels else nn.Identity(),
-            DuplicateChannels() if duplicate_channels else nn.Identity(),
+            Stereo() if use_stereo else nn.Identity(),
             Loudness(sampling_rate=target_rate, target=loudness)  # type: ignore
             if exists(loudness)
             else nn.Identity(),

audio_data_pytorch/transforms/stereo.py

@@ -0,0 +1,15 @@
+from torch import Tensor, nn
+
+
+class Stereo(nn.Module):
+    def forward(self, x: Tensor) -> Tensor:
+        shape = x.shape
+        channels = shape[0]
+        if len(shape) == 1:  # s -> 2, s
+            x = x.unsqueeze(0).repeat(2, 1)
+        elif len(shape) == 2:
+            if channels == 1:  # 1, s -> 2, s
+                x = x.repeat(2, 1)
+            elif channels > 2:  # ?, s -> 2,s
+                x = x[:2, :]
+        return x

setup.py

@@ -3,7 +3,7 @@
 setup(
     name="audio-data-pytorch",
     packages=find_packages(exclude=[]),
-    version="0.0.8",
+    version="0.0.9",
     license="MIT",
     description="Audio Data - PyTorch",
     long_description_content_type="text/markdown",