From 3ec92c5bd5d4b16853774cf1ed620dccc75e89ca Mon Sep 17 00:00:00 2001
From: pseeth <prem@u.northwestern.edu>
Date: Mon, 13 Feb 2023 19:08:43 -0600
Subject: [PATCH] Padding amount was too high for loading files with FFMPEG.
 (#72)

* Getting the pad amount as low as possible for mp3 to pass.

* Fleshing out comment.

* Bumping version

* Raising threshold.

* Adding source

* updating comment

* Adding offset option to dataset

* Changing to hann window

* Update regression data.

* Raising atol but why

---------

Co-authored-by: pseeth <prem@descript.com>
---
 .gitignore                                       |  1 +
 .pre-commit-config.yaml                          |  2 +-
 audiotools/__init__.py                           |  2 +-
 audiotools/core/audio_signal.py                  |  2 +-
 audiotools/core/ffmpeg.py                        |  8 ++++++--
 audiotools/data/datasets.py                      |  4 +++-
 audiotools/data/transforms.py                    |  2 --
 audiotools/metrics/spectral.py                   | 16 ++++++++++++++--
 audiotools/ml/layers/spectral_gate.py            |  1 -
 audiotools/post.py                               |  1 -
 setup.py                                         |  2 +-
 tests/core/test_grad.py                          |  4 ----
 tests/data/test_transforms.py                    |  2 +-
 tests/regression/transforms/BackgroundNoise.wav  |  4 ++--
 tests/regression/transforms/BaseTransform.wav    |  4 ++--
 tests/regression/transforms/Choose.wav           |  4 ++--
 .../regression/transforms/ClippingDistortion.wav |  4 ++--
 tests/regression/transforms/Compose.wav          |  4 ++--
 tests/regression/transforms/CorruptPhase.wav     |  4 ++--
 tests/regression/transforms/CrossTalk.wav        |  4 ++--
 tests/regression/transforms/Equalizer.wav        |  4 ++--
 tests/regression/transforms/FrequencyMask.wav    |  4 ++--
 tests/regression/transforms/FrequencyNoise.wav   |  4 ++--
 tests/regression/transforms/GlobalVolumeNorm.wav |  4 ++--
 tests/regression/transforms/HighPass.wav         |  4 ++--
 tests/regression/transforms/Identity.wav         |  4 ++--
 tests/regression/transforms/InvertPhase.wav      |  4 ++--
 tests/regression/transforms/LowPass.wav          |  4 ++--
 .../regression/transforms/MaskLowMagnitudes.wav  |  4 ++--
 .../regression/transforms/MuLawQuantization.wav  |  4 ++--
 tests/regression/transforms/NoiseFloor.wav       |  4 ++--
 tests/regression/transforms/Quantization.wav     |  4 ++--
 tests/regression/transforms/Repeat.wav           |  4 ++--
 tests/regression/transforms/RepeatUpTo.wav       |  4 ++--
 tests/regression/transforms/RescaleAudio.wav     |  4 ++--
 .../transforms/RoomImpulseResponse.wav           |  4 ++--
 tests/regression/transforms/ShiftPhase.wav       |  4 ++--
 tests/regression/transforms/Silence.wav          |  4 ++--
 tests/regression/transforms/Smoothing.wav        |  4 ++--
 .../regression/transforms/SpectralDenoising.wav  |  4 ++--
 .../regression/transforms/SpectralTransform.wav  |  4 ++--
 tests/regression/transforms/TimeMask.wav         |  4 ++--
 tests/regression/transforms/TimeNoise.wav        |  4 ++--
 tests/regression/transforms/VolumeChange.wav     |  4 ++--
 tests/regression/transforms/VolumeNorm.wav       |  4 ++--
 45 files changed, 93 insertions(+), 82 deletions(-)

diff --git a/.gitignore b/.gitignore
index 89c64ec7..381833bb 100644
--- a/.gitignore
+++ b/.gitignore
@@ -140,3 +140,4 @@ test_out.md
 
 # generated test data
 tests/audio/chords/
+notebooks/*
diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
index 589edc7e..eca3cf91 100644
--- a/.pre-commit-config.yaml
+++ b/.pre-commit-config.yaml
@@ -4,7 +4,7 @@ repos:
   hooks:
     - id: reorder-python-imports
 - repo: https://github.com/psf/black
-  rev: 22.3.0
+  rev: 23.1.0
   hooks:
     - id: black
       language_version: python3
diff --git a/audiotools/__init__.py b/audiotools/__init__.py
index c4945117..945edf28 100644
--- a/audiotools/__init__.py
+++ b/audiotools/__init__.py
@@ -1,4 +1,4 @@
-__version__ = "0.6.0"
+__version__ = "0.6.1"
 from .core import AudioSignal
 from .core import STFTParams
 from .core import Meter
diff --git a/audiotools/core/audio_signal.py b/audiotools/core/audio_signal.py
index 6ad25a85..3975fa71 100644
--- a/audiotools/core/audio_signal.py
+++ b/audiotools/core/audio_signal.py
@@ -1063,7 +1063,7 @@ def stft_params(self):
     def stft_params(self, value: STFTParams):
         default_win_len = int(2 ** (np.ceil(np.log2(0.032 * self.sample_rate))))
         default_hop_len = default_win_len // 4
-        default_win_type = "sqrt_hann"
+        default_win_type = "hann"
         default_match_stride = False
         default_padding_type = "reflect"
 
diff --git a/audiotools/core/ffmpeg.py b/audiotools/core/ffmpeg.py
index 7d16e059..134b4ca5 100644
--- a/audiotools/core/ffmpeg.py
+++ b/audiotools/core/ffmpeg.py
@@ -178,8 +178,12 @@ def load_from_file_with_ffmpeg(cls, audio_path: str, quiet: bool = True, **kwarg
             # offset in a video container.
             pad = ffprobe_offset(audio_path)
             # Don't pad files with discrepancies less than
-            # 0.1s - it's likely due to codec latency.
-            if pad < 0.1:
+            # 0.0.27s - it's likely due to codec latency.
+            # The amount of latency introduced by mp3 is
+            # 1152, which is 0.0261 44khz. So we
+            # set the threshold here slightly above that.
+            # Source: https://lame.sourceforge.io/tech-FAQ.txt.
+            if pad < 0.027:
                 pad = 0.0
             ff = ffmpy.FFmpeg(
                 inputs={wav_file: None},
diff --git a/audiotools/data/datasets.py b/audiotools/data/datasets.py
index b0208eeb..cad2be5c 100644
--- a/audiotools/data/datasets.py
+++ b/audiotools/data/datasets.py
@@ -334,6 +334,7 @@ def __init__(
         sample_rate: int,
         n_examples: int = 1000,
         duration: float = 0.5,
+        offset: float = None,
         loudness_cutoff: float = -40,
         num_channels: int = 1,
         transform: Callable = None,
@@ -355,6 +356,7 @@ def __init__(
         self.transform = transform
         self.sample_rate = sample_rate
         self.duration = duration
+        self.offset = offset
         self.aligned = aligned
         self.shuffle_loaders = shuffle_loaders
 
@@ -367,7 +369,7 @@ def __init__(
 
     def __getitem__(self, idx):
         state = util.random_state(idx)
-        offset = None
+        offset = None if self.offset is None else self.offset
         item = {}
 
         keys = list(self.loaders.keys())
diff --git a/audiotools/data/transforms.py b/audiotools/data/transforms.py
index 20d301e2..504e87dc 100644
--- a/audiotools/data/transforms.py
+++ b/audiotools/data/transforms.py
@@ -493,7 +493,6 @@ def __init__(
         name: str = None,
         prob: float = 1.0,
     ):
-
         transforms = [copy.copy(transform) for _ in range(n_repeat)]
         super().__init__(transforms, name=name, prob=prob)
 
@@ -1479,7 +1478,6 @@ def __init__(
         name: str = None,
         prob: float = 1,
     ):
-
         super().__init__(t_center=t_center, t_width=t_width, name=name, prob=prob)
 
     def _transform(self, signal, tmin_s: float, tmax_s: float):
diff --git a/audiotools/metrics/spectral.py b/audiotools/metrics/spectral.py
index 979256cc..7ce95388 100644
--- a/audiotools/metrics/spectral.py
+++ b/audiotools/metrics/spectral.py
@@ -48,10 +48,16 @@ def __init__(
         pow: float = 2.0,
         weight: float = 1.0,
         match_stride: bool = False,
+        window_type: str = None,
     ):
         super().__init__()
         self.stft_params = [
-            STFTParams(window_length=w, hop_length=w // 4, match_stride=match_stride)
+            STFTParams(
+                window_length=w,
+                hop_length=w // 4,
+                match_stride=match_stride,
+                window_type=window_type,
+            )
             for w in window_lengths
         ]
         self.loss_fn = loss_fn
@@ -128,10 +134,16 @@ def __init__(
         match_stride: bool = False,
         mel_fmin: List[float] = [0.0, 0.0],
         mel_fmax: List[float] = [None, None],
+        window_type: str = None,
     ):
         super().__init__()
         self.stft_params = [
-            STFTParams(window_length=w, hop_length=w // 4, match_stride=match_stride)
+            STFTParams(
+                window_length=w,
+                hop_length=w // 4,
+                match_stride=match_stride,
+                window_type=window_type,
+            )
             for w in window_lengths
         ]
         self.n_mels = n_mels
diff --git a/audiotools/ml/layers/spectral_gate.py b/audiotools/ml/layers/spectral_gate.py
index cc39746c..c4ae8b5e 100644
--- a/audiotools/ml/layers/spectral_gate.py
+++ b/audiotools/ml/layers/spectral_gate.py
@@ -35,7 +35,6 @@ class SpectralGate(nn.Module):
     """
 
     def __init__(self, n_freq: int = 3, n_time: int = 5):
-
         super().__init__()
 
         smoothing_filter = torch.outer(
diff --git a/audiotools/post.py b/audiotools/post.py
index 9be24720..45903fea 100644
--- a/audiotools/post.py
+++ b/audiotools/post.py
@@ -111,7 +111,6 @@ def _default_format_fn(label, x, **kwargs):
 
         v_keys = list(v.keys())
         if columns is None:
-
             columns = [first_column] + v_keys
             output.append(" | ".join(columns))
 
diff --git a/setup.py b/setup.py
index dee54b67..4619be9a 100644
--- a/setup.py
+++ b/setup.py
@@ -6,7 +6,7 @@
 
 setup(
     name="audiotools",
-    version="0.6.0",
+    version="0.6.1",
     classifiers=[
         "Intended Audience :: Developers",
         "Intended Audience :: Education",
diff --git a/tests/core/test_grad.py b/tests/core/test_grad.py
index b3761548..ee27d248 100644
--- a/tests/core/test_grad.py
+++ b/tests/core/test_grad.py
@@ -9,12 +9,10 @@
 
 
 def test_audio_grad():
-
     audio_path = "tests/audio/spk/f10_script4_produced.wav"
     ir_path = "tests/audio/ir/h179_Bar_1txts.wav"
 
     def _test_audio_grad(attr: str, target=True, kwargs: dict = {}):
-
         signal = AudioSignal(audio_path)
         signal.audio_data.requires_grad = True
 
@@ -26,7 +24,6 @@ def _test_audio_grad(attr: str, target=True, kwargs: dict = {}):
 
         try:
             if isinstance(result, AudioSignal):
-
                 # If necessary, propagate spectrogram changes to waveform
                 if result.stft_data is not None:
                     result.istft()
@@ -88,7 +85,6 @@ def _test_audio_grad(attr: str, target=True, kwargs: dict = {}):
 
 
 def test_batch_grad():
-
     audio_path = "tests/audio/spk/f10_script4_produced.wav"
 
     signal = AudioSignal(audio_path)
diff --git a/tests/data/test_transforms.py b/tests/data/test_transforms.py
index 02d709b4..f46a1b0f 100644
--- a/tests/data/test_transforms.py
+++ b/tests/data/test_transforms.py
@@ -25,7 +25,7 @@ def _compare_transform(transform_name, signal):
     if regression_data.exists():
         regression_signal = AudioSignal(regression_data)
         assert torch.allclose(
-            signal.audio_data, regression_signal.audio_data, atol=1e-6
+            signal.audio_data, regression_signal.audio_data, atol=1e-4
         )
     else:
         signal.write(regression_data)
diff --git a/tests/regression/transforms/BackgroundNoise.wav b/tests/regression/transforms/BackgroundNoise.wav
index c175a010..daf9cce1 100644
--- a/tests/regression/transforms/BackgroundNoise.wav
+++ b/tests/regression/transforms/BackgroundNoise.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:be582a2a14d38a5652d93cf2ef461e782957090765fb90d4e1c0345009c87b44
-size 352858
+oid sha256:3d54d33d4ebfa647cd5cfbd0b8c7b9333fc6cce3b0bba7c127373de19a276cb6
+size 176444
diff --git a/tests/regression/transforms/BaseTransform.wav b/tests/regression/transforms/BaseTransform.wav
index 9ccb1ef4..864c96bd 100644
--- a/tests/regression/transforms/BaseTransform.wav
+++ b/tests/regression/transforms/BaseTransform.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ff8e9603e10a47725204417d22e40e775cccbfd033445196f5ae313a3a27a451
-size 352858
+oid sha256:4ae9b85271fcc0cdda6ec81cc9e119179647755763d1f12e7015333547d053d9
+size 176444
diff --git a/tests/regression/transforms/Choose.wav b/tests/regression/transforms/Choose.wav
index 4dc5935e..0d87aa11 100644
--- a/tests/regression/transforms/Choose.wav
+++ b/tests/regression/transforms/Choose.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8dd119b24b48b9d668b27045fa36bf5dfdecd7523df139709fd371444e0f09a0
-size 352858
+oid sha256:c5bebfd1bd6ec8ab0eb26dcef0cab1b68effb11de80538ecd34bf788e501fd00
+size 176444
diff --git a/tests/regression/transforms/ClippingDistortion.wav b/tests/regression/transforms/ClippingDistortion.wav
index f745961f..3f244af9 100644
--- a/tests/regression/transforms/ClippingDistortion.wav
+++ b/tests/regression/transforms/ClippingDistortion.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:46ecacdd41aead53e21a6a3d9a3c88ba052d52f27dc10035596d91f12f19b28d
-size 352858
+oid sha256:373df372149380e97f1c87cd8b65749792da023ff7828229626f59dc225622ad
+size 176444
diff --git a/tests/regression/transforms/Compose.wav b/tests/regression/transforms/Compose.wav
index 2be53419..a62f0343 100644
--- a/tests/regression/transforms/Compose.wav
+++ b/tests/regression/transforms/Compose.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:eda39f01db59569b885ddd0a1b34fa6c1135201fcff3bc95f0ba83a03521182a
-size 352858
+oid sha256:4e67c650af21c88b01c1ddb5d443abef9719d373f2d434973a5a29cab8609ec2
+size 176444
diff --git a/tests/regression/transforms/CorruptPhase.wav b/tests/regression/transforms/CorruptPhase.wav
index 0b03f31e..9c8cbd80 100644
--- a/tests/regression/transforms/CorruptPhase.wav
+++ b/tests/regression/transforms/CorruptPhase.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:53b6bad6816b44a04568a1bf0d73d1d471dbddbb1ce5ee8750e86dd261e5054e
-size 352858
+oid sha256:46f911803d68580b038b2d6b5b91dd4e07e5ccb5bd1b1046bfff4c4234f2f180
+size 176444
diff --git a/tests/regression/transforms/CrossTalk.wav b/tests/regression/transforms/CrossTalk.wav
index 62ffe8a2..06ada81e 100644
--- a/tests/regression/transforms/CrossTalk.wav
+++ b/tests/regression/transforms/CrossTalk.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f201403e2142e29ed6834221ed10d96a67365458e640c08175dd8571177a2908
-size 352858
+oid sha256:718ab88ce7cc2a1e184f79d18c5c45a254900a0713a8c855f6ebe892b400d1f9
+size 176444
diff --git a/tests/regression/transforms/Equalizer.wav b/tests/regression/transforms/Equalizer.wav
index ebf8b70d..1b2bd930 100644
--- a/tests/regression/transforms/Equalizer.wav
+++ b/tests/regression/transforms/Equalizer.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3e8773268ce7c5a261a83ae095b279e85189a771d7e845aea981fe771a3c63fb
-size 352858
+oid sha256:5fb5e279ead5519f96b55e635978dfdca00409ef7804333c230beedad2a79623
+size 176444
diff --git a/tests/regression/transforms/FrequencyMask.wav b/tests/regression/transforms/FrequencyMask.wav
index 0db44266..283ed7f9 100644
--- a/tests/regression/transforms/FrequencyMask.wav
+++ b/tests/regression/transforms/FrequencyMask.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8268d28cbb5f6b99f68fd76ccc25d5f23370dd1b4e125271cbd93f74e192a6a7
-size 352858
+oid sha256:be3c30adb4801f599771fa1dcb5f59c1630edb5cc7f78ee175781284914c3b02
+size 176444
diff --git a/tests/regression/transforms/FrequencyNoise.wav b/tests/regression/transforms/FrequencyNoise.wav
index 1f3e8de8..06adc439 100644
--- a/tests/regression/transforms/FrequencyNoise.wav
+++ b/tests/regression/transforms/FrequencyNoise.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a55fa29aa9da395760d7c2b3fd10a220f103f6410a2c65779c86001242e8cb7b
-size 352858
+oid sha256:19ea8e2f05a8bf0564c85623ab1b54822bdacff64fbbb6b773d74d8e267db992
+size 176444
diff --git a/tests/regression/transforms/GlobalVolumeNorm.wav b/tests/regression/transforms/GlobalVolumeNorm.wav
index e963c5ee..17fb6e57 100644
--- a/tests/regression/transforms/GlobalVolumeNorm.wav
+++ b/tests/regression/transforms/GlobalVolumeNorm.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a1b3416db578cb541a5cb360c22e2b0a36fd3a57d644b7ce8b804108a3f84510
-size 352858
+oid sha256:cd1d2949a6be1d6bd08a787ffba8363c68ddcc36e3a73110e026088f8b8c1b6f
+size 176444
diff --git a/tests/regression/transforms/HighPass.wav b/tests/regression/transforms/HighPass.wav
index aa80da5d..dd831d7a 100644
--- a/tests/regression/transforms/HighPass.wav
+++ b/tests/regression/transforms/HighPass.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a85d10f21ba799f6125b59351fa512ae030aa68380fc754fc29fbaeade68dbbd
-size 352858
+oid sha256:be66e2972da9dc2fee575cd36b0b153c1abd89cfbd05d9e221b506bba1677575
+size 176444
diff --git a/tests/regression/transforms/Identity.wav b/tests/regression/transforms/Identity.wav
index 9ccb1ef4..864c96bd 100644
--- a/tests/regression/transforms/Identity.wav
+++ b/tests/regression/transforms/Identity.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ff8e9603e10a47725204417d22e40e775cccbfd033445196f5ae313a3a27a451
-size 352858
+oid sha256:4ae9b85271fcc0cdda6ec81cc9e119179647755763d1f12e7015333547d053d9
+size 176444
diff --git a/tests/regression/transforms/InvertPhase.wav b/tests/regression/transforms/InvertPhase.wav
index a2104d6f..24d09766 100644
--- a/tests/regression/transforms/InvertPhase.wav
+++ b/tests/regression/transforms/InvertPhase.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:65f7c3fd91674ef04c1b208427055a112d562ae172360ecccd03388b2d0c1275
-size 352858
+oid sha256:9fe26073ef22c08dcffa45731d86bd17b3e1bae3613e81889d137b8ff60c1cb1
+size 176444
diff --git a/tests/regression/transforms/LowPass.wav b/tests/regression/transforms/LowPass.wav
index 0f7a2770..9592447b 100644
--- a/tests/regression/transforms/LowPass.wav
+++ b/tests/regression/transforms/LowPass.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4229ed6c1139b9b8b00bc79397f3f300583581380b8a4226eb9eb5b4d7d7bc13
-size 352858
+oid sha256:0e965b9dd27fa2fcde431c7be9bbb0087111a954cec67dd28060c9a0cb10326f
+size 176444
diff --git a/tests/regression/transforms/MaskLowMagnitudes.wav b/tests/regression/transforms/MaskLowMagnitudes.wav
index a434ca75..7b1efa85 100644
--- a/tests/regression/transforms/MaskLowMagnitudes.wav
+++ b/tests/regression/transforms/MaskLowMagnitudes.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c171ca488f894bbd90482201c73efd7ed5faf9a796adaacda3e8f37261799148
-size 352858
+oid sha256:cf442aa2b861653b0388d18c7b4cd54cb8554dcef2d9fb1f4ab755d564d4c465
+size 176444
diff --git a/tests/regression/transforms/MuLawQuantization.wav b/tests/regression/transforms/MuLawQuantization.wav
index e41d5fbf..5b2317de 100644
--- a/tests/regression/transforms/MuLawQuantization.wav
+++ b/tests/regression/transforms/MuLawQuantization.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:db75b51697f7b4da2ac5fa73088d1a9397f72cb8f899c6f5affc1c632f1f52c0
-size 352858
+oid sha256:4759c209a9d64a1ea33ddbf4365f704751913538c4d6b26f715ce5b2a5ae7ac6
+size 176444
diff --git a/tests/regression/transforms/NoiseFloor.wav b/tests/regression/transforms/NoiseFloor.wav
index 86967f15..b3a3e9db 100644
--- a/tests/regression/transforms/NoiseFloor.wav
+++ b/tests/regression/transforms/NoiseFloor.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c45db88d199526a9299e906a0ddfb25458403d4c95715f7535d2e68bc35c54db
-size 352858
+oid sha256:bec9c93df61018139f76d3f480d653a42d4c92d1b10e513efc48a7041be7b0d4
+size 176444
diff --git a/tests/regression/transforms/Quantization.wav b/tests/regression/transforms/Quantization.wav
index f7c6fd04..f6813b82 100644
--- a/tests/regression/transforms/Quantization.wav
+++ b/tests/regression/transforms/Quantization.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:71a1131776ef33f1763a104fff27bd4d1b7add27858162de04e73004b6cc01be
-size 352858
+oid sha256:9361bc6f3a13e94fef422cf416f98fd71fe5a211bb649ce30eb5877865ad1449
+size 176444
diff --git a/tests/regression/transforms/Repeat.wav b/tests/regression/transforms/Repeat.wav
index d2a114e5..b87a407c 100644
--- a/tests/regression/transforms/Repeat.wav
+++ b/tests/regression/transforms/Repeat.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bfacc1f593e8cacf6329e52fb31dee4cd6f9b396f2adf88fb7727bb363d54501
-size 352858
+oid sha256:ea4d8f7a3cd040fd436209169943aa16c85666a8736631c4a4e96626ecde79d2
+size 176444
diff --git a/tests/regression/transforms/RepeatUpTo.wav b/tests/regression/transforms/RepeatUpTo.wav
index 2de5e515..5f975ef8 100644
--- a/tests/regression/transforms/RepeatUpTo.wav
+++ b/tests/regression/transforms/RepeatUpTo.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:aa5a061288cc7de5f59d1924d9a42e731062302b85121391becc3ef2b35b676d
-size 352858
+oid sha256:7e4f826c181a383154cdd4cd6208738ac6c044be59f3f0c1c00167fc473c3068
+size 176444
diff --git a/tests/regression/transforms/RescaleAudio.wav b/tests/regression/transforms/RescaleAudio.wav
index 9ccb1ef4..864c96bd 100644
--- a/tests/regression/transforms/RescaleAudio.wav
+++ b/tests/regression/transforms/RescaleAudio.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ff8e9603e10a47725204417d22e40e775cccbfd033445196f5ae313a3a27a451
-size 352858
+oid sha256:4ae9b85271fcc0cdda6ec81cc9e119179647755763d1f12e7015333547d053d9
+size 176444
diff --git a/tests/regression/transforms/RoomImpulseResponse.wav b/tests/regression/transforms/RoomImpulseResponse.wav
index 4dc5935e..0d87aa11 100644
--- a/tests/regression/transforms/RoomImpulseResponse.wav
+++ b/tests/regression/transforms/RoomImpulseResponse.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8dd119b24b48b9d668b27045fa36bf5dfdecd7523df139709fd371444e0f09a0
-size 352858
+oid sha256:c5bebfd1bd6ec8ab0eb26dcef0cab1b68effb11de80538ecd34bf788e501fd00
+size 176444
diff --git a/tests/regression/transforms/ShiftPhase.wav b/tests/regression/transforms/ShiftPhase.wav
index 54f79bda..6d47cf21 100644
--- a/tests/regression/transforms/ShiftPhase.wav
+++ b/tests/regression/transforms/ShiftPhase.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:203d0327eca69578652651f1ad87e12036226289d2c9e0438084a53e7a6d18a2
-size 352858
+oid sha256:4708845550eed14fc69a125a5ef3a26953afd8438101d4f1862dfbf2c5378935
+size 176444
diff --git a/tests/regression/transforms/Silence.wav b/tests/regression/transforms/Silence.wav
index fc5219fc..14b225a2 100644
--- a/tests/regression/transforms/Silence.wav
+++ b/tests/regression/transforms/Silence.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a7dcd5d95cb0896262a742cc64124737b3e94529c2c38753bdd681f282cc9f35
-size 352858
+oid sha256:cfc6b206e99b298a229480f020e61d2b22a791e08a840fdd40ef62d8bd78b155
+size 176444
diff --git a/tests/regression/transforms/Smoothing.wav b/tests/regression/transforms/Smoothing.wav
index 209920c5..838eb3e9 100644
--- a/tests/regression/transforms/Smoothing.wav
+++ b/tests/regression/transforms/Smoothing.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c3d3ae2963e4969d52a1eacfd767f30d7b7f061fe400e5addf8e1de01cb452f4
-size 352858
+oid sha256:cd1ee5b00278f342d5926ad9f9b662cdf2176f2f17a626d4ee8f5a37393ca7b7
+size 176444
diff --git a/tests/regression/transforms/SpectralDenoising.wav b/tests/regression/transforms/SpectralDenoising.wav
index 975afeae..2412d25e 100644
--- a/tests/regression/transforms/SpectralDenoising.wav
+++ b/tests/regression/transforms/SpectralDenoising.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5d7de336bfc6f7f9dddec46bdbb878c2d61b0bff43796f8f1c858e0173e35299
-size 352858
+oid sha256:ba7fdc5bff221a8c479cc1c5db598d91d02e26b0b4544948f406729b2b4a58b2
+size 176444
diff --git a/tests/regression/transforms/SpectralTransform.wav b/tests/regression/transforms/SpectralTransform.wav
index b972c1a4..30161647 100644
--- a/tests/regression/transforms/SpectralTransform.wav
+++ b/tests/regression/transforms/SpectralTransform.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:feb1a3000ae1f6e9cb7dd2c82c2bb42bf3b1f7dcc6d2f124b97b348392a12a4f
-size 352858
+oid sha256:2e294a2d121916653e64c811d71dc07faefa8b2c8c55fba390efa511b254c836
+size 176444
diff --git a/tests/regression/transforms/TimeMask.wav b/tests/regression/transforms/TimeMask.wav
index 9bb97e1c..6cccd25b 100644
--- a/tests/regression/transforms/TimeMask.wav
+++ b/tests/regression/transforms/TimeMask.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cc7922d9dba0c31433a809277520b213463da6792138406d121c2334107f165a
-size 352858
+oid sha256:1b591bf382c79f7887b3ea3ffbb1a3c9b79e8290aeccd056c07e1ff4d8730e4d
+size 176444
diff --git a/tests/regression/transforms/TimeNoise.wav b/tests/regression/transforms/TimeNoise.wav
index 63b471e2..09c966c3 100644
--- a/tests/regression/transforms/TimeNoise.wav
+++ b/tests/regression/transforms/TimeNoise.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c0d69ba9f4ad09ccf60fe95343078ab0df48ed70875a27d5c599903b4e8df56f
-size 352858
+oid sha256:b50b14ed6aa3a878398654654da5ef6fb724e75cf9b529e092dfa1c1db535962
+size 176444
diff --git a/tests/regression/transforms/VolumeChange.wav b/tests/regression/transforms/VolumeChange.wav
index 641fc1ba..214e333f 100644
--- a/tests/regression/transforms/VolumeChange.wav
+++ b/tests/regression/transforms/VolumeChange.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6f0ab8a1d888f05b591a5a2b59f2a623d1588162cde1733372c86b37bfa257ad
-size 352858
+oid sha256:1ff621636bc3199ea91ef342df787bf20fce7ee1a5a879080b59b036e0ed985a
+size 176444
diff --git a/tests/regression/transforms/VolumeNorm.wav b/tests/regression/transforms/VolumeNorm.wav
index 2c538385..d589b88b 100644
--- a/tests/regression/transforms/VolumeNorm.wav
+++ b/tests/regression/transforms/VolumeNorm.wav
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:86538fbb9a4b749caf414888ea8d98989720e3129505f7c9e5067143aeb5509e
-size 352858
+oid sha256:cd6417903e035d2a0838d3ee550c626fbc228934a12bfd3e1ecf08dc7019713b
+size 176444