add st-tio module

st_ito/LICENSE

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

st_ito/models/panns.py

@@ -0,0 +1,281 @@
+# Adapted from https://github.com/qiuqiangkong/audioset_tagging_cnn/blob/master/pytorch/models.py
+# Under MIT License
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from torchlibrosa.stft import Spectrogram, LogmelFilterBank
+from torchlibrosa.augmentation import SpecAugmentation
+
+
+def init_layer(layer):
+    """Initialize a Linear or Convolutional layer."""
+    nn.init.xavier_uniform_(layer.weight)
+
+    if hasattr(layer, "bias"):
+        if layer.bias is not None:
+            layer.bias.data.fill_(0.0)
+
+
+def init_bn(bn):
+    """Initialize a Batchnorm layer."""
+    bn.bias.data.fill_(0.0)
+    bn.weight.data.fill_(1.0)
+
+
+class ConvBlock(nn.Module):
+    def __init__(self, in_channels, out_channels, use_batchnorm: bool = True):
+        super(ConvBlock, self).__init__()
+        self.use_batchnorm = use_batchnorm
+
+        self.conv1 = nn.Conv2d(
+            in_channels=in_channels,
+            out_channels=out_channels,
+            kernel_size=(3, 3),
+            stride=(1, 1),
+            padding=(1, 1),
+            bias=False,
+        )
+
+        self.conv2 = nn.Conv2d(
+            in_channels=out_channels,
+            out_channels=out_channels,
+            kernel_size=(3, 3),
+            stride=(1, 1),
+            padding=(1, 1),
+            bias=False,
+        )
+
+        if self.use_batchnorm:
+            self.bn1 = nn.BatchNorm2d(out_channels)
+            self.bn2 = nn.BatchNorm2d(out_channels)
+        else:
+            self.bn1 = nn.Identity()
+            self.bn2 = nn.Identity()
+
+        self.init_weight()
+
+    def init_weight(self):
+        init_layer(self.conv1)
+        init_layer(self.conv2)
+
+        if self.use_batchnorm:
+            init_bn(self.bn1)
+            init_bn(self.bn2)
+
+    def forward(self, input, pool_size=(2, 2), pool_type="avg"):
+        x = input
+        x = F.relu_(self.bn1(self.conv1(x)))
+        x = F.relu_(self.bn2(self.conv2(x)))
+        if pool_type == "max":
+            x = F.max_pool2d(x, kernel_size=pool_size)
+        elif pool_type == "avg":
+            x = F.avg_pool2d(x, kernel_size=pool_size)
+        elif pool_type == "avg+max":
+            x1 = F.avg_pool2d(x, kernel_size=pool_size)
+            x2 = F.max_pool2d(x, kernel_size=pool_size)
+            x = x1 + x2
+        else:
+            raise Exception("Incorrect argument!")
+
+        return x
+
+
+class ConvBlock5x5(nn.Module):
+    def __init__(self, in_channels, out_channels):
+        super(ConvBlock5x5, self).__init__()
+
+        self.conv1 = nn.Conv2d(
+            in_channels=in_channels,
+            out_channels=out_channels,
+            kernel_size=(5, 5),
+            stride=(1, 1),
+            padding=(2, 2),
+            bias=False,
+        )
+
+        self.bn1 = nn.BatchNorm2d(out_channels)
+
+        self.init_weight()
+
+    def init_weight(self):
+        init_layer(self.conv1)
+        init_bn(self.bn1)
+
+    def forward(self, input, pool_size=(2, 2), pool_type="avg"):
+        x = input
+        x = F.relu_(self.bn1(self.conv1(x)))
+        if pool_type == "max":
+            x = F.max_pool2d(x, kernel_size=pool_size)
+        elif pool_type == "avg":
+            x = F.avg_pool2d(x, kernel_size=pool_size)
+        elif pool_type == "avg+max":
+            x1 = F.avg_pool2d(x, kernel_size=pool_size)
+            x2 = F.max_pool2d(x, kernel_size=pool_size)
+            x = x1 + x2
+        else:
+            raise Exception("Incorrect argument!")
+
+        return x
+
+
+class Cnn14(nn.Module):
+    def __init__(
+        self,
+        embed_dim: int,
+        sample_rate: float,
+        window_size: int,
+        hop_size: int,
+        mel_bins: int,
+        fmin: float,
+        fmax: float,
+        use_batchnorm: bool = False,
+        input_norm: str = "batchnorm",
+    ):
+        super(Cnn14, self).__init__()
+        self.embed_dim = embed_dim
+        self.use_batchnorm = use_batchnorm
+        self.input_norm = input_norm
+
+        window = "hann"
+        center = True
+        pad_mode = "reflect"
+        ref = 1.0
+        amin = 1e-10
+        top_db = None
+
+        # Spectrogram extractor
+        self.spectrogram_extractor = Spectrogram(
+            n_fft=window_size,
+            hop_length=hop_size,
+            win_length=window_size,
+            window=window,
+            center=center,
+            pad_mode=pad_mode,
+            freeze_parameters=True,
+        )
+
+        # Logmel feature extractor
+        self.logmel_extractor = LogmelFilterBank(
+            sr=sample_rate,
+            n_fft=window_size,
+            n_mels=mel_bins,
+            fmin=fmin,
+            fmax=fmax,
+            ref=ref,
+            amin=amin,
+            top_db=top_db,
+            freeze_parameters=True,
+        )
+
+        # Spec augmenter
+        self.spec_augmenter = SpecAugmentation(
+            time_drop_width=64,
+            time_stripes_num=2,
+            freq_drop_width=8,
+            freq_stripes_num=2,
+        )
+
+        self.bn0 = nn.BatchNorm2d(mel_bins)
+
+        self.conv_block1 = ConvBlock(
+            in_channels=1, out_channels=64, use_batchnorm=use_batchnorm
+        )
+        self.conv_block2 = ConvBlock(
+            in_channels=64, out_channels=128, use_batchnorm=use_batchnorm
+        )
+        self.conv_block3 = ConvBlock(
+            in_channels=128, out_channels=256, use_batchnorm=use_batchnorm
+        )
+        self.conv_block4 = ConvBlock(
+            in_channels=256, out_channels=512, use_batchnorm=use_batchnorm
+        )
+        self.conv_block5 = ConvBlock(
+            in_channels=512, out_channels=1024, use_batchnorm=use_batchnorm
+        )
+        self.conv_block6 = ConvBlock(
+            in_channels=1024, out_channels=2048, use_batchnorm=use_batchnorm
+        )
+
+        self.fc_mid = nn.Linear(2048, embed_dim, bias=True)
+        self.fc_side = nn.Linear(2048, embed_dim, bias=True)
+
+        self.init_weight()
+
+    def init_weight(self):
+        init_bn(self.bn0)
+        init_layer(self.fc_mid)
+        init_layer(self.fc_side)
+
+    def forward(self, x: torch.Tensor):
+        """
+        input (torch.Tensor): Waveform tensor with shape (batch_size, chs, seq_len)
+        """
+        batch_size, chs, seq_len = x.size()
+
+        # compute mid and side signals
+        if chs == 1:
+            pass
+        elif chs == 2:
+            x_mid = (x[:, 0, :] + x[:, 1, :]) / 2
+            x_side = (x[:, 0, :] - x[:, 1, :]) / 2
+            # stack along batch dim
+            x = torch.stack([x_mid, x_side], dim=1)
+        else:
+            raise ValueError(f"Invalid number of channels: {chs}")
+
+        # move to batch dim
+        x = x.view(batch_size * chs, seq_len)
+
+        # extract logmel features
+        x = self.spectrogram_extractor(x)  # (batch_size, 1, time_steps, freq_bins)
+        x = self.logmel_extractor(x)  # (batch_size, 1, time_steps, mel_bins)
+
+        if self.input_norm == "batchnorm":
+            # this normalizes over mel bins which is problematic for equalization
+            x = x.transpose(1, 3)
+            x = self.bn0(x)
+            x = x.transpose(1, 3)
+        elif self.input_norm == "minmax":
+            x = x.clamp(-80, 40.0)  # clamp the logmels between -80 and 40
+            x = (x + 80) / 120  # normalize the logmels between 0 and 1
+            x = (x * 2) - 1  # normalize the logmels between -1 and 1
+        elif self.input_norm == "none":
+            pass
+        else:
+            raise ValueError(f"Invalid input_norm: {self.input_norm}")
+
+        if self.training:
+            x = self.spec_augmenter(x)
+
+        x = self.conv_block1(x, pool_size=(2, 2), pool_type="avg")
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = self.conv_block2(x, pool_size=(2, 2), pool_type="avg")
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = self.conv_block3(x, pool_size=(2, 2), pool_type="avg")
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = self.conv_block4(x, pool_size=(2, 2), pool_type="avg")
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = self.conv_block5(x, pool_size=(2, 2), pool_type="avg")
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = self.conv_block6(x, pool_size=(1, 1), pool_type="avg")
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = torch.mean(x, dim=3)
+
+        (x1, _) = torch.max(x, dim=2)
+        x2 = torch.mean(x, dim=2)
+        x = x1 + x2
+
+        # move mid and side back to channel dim
+        x = x.view(batch_size, chs, -1)
+
+        if chs == 1:
+            x_mid = x[:, 0, :]
+            mid_embed = self.fc_mid(x_mid)
+            side_embed = mid_embed
+        elif chs == 2:
+            x_mid = x[:, 0, :]
+            x_side = x[:, 1, :]
+            mid_embed = self.fc_mid(x_mid)
+            side_embed = self.fc_side(x_side)
+
+        return mid_embed, side_embed

st_ito/utils.py

@@ -0,0 +1,222 @@
+import os
+import yaml
+import torch
+import torchaudio
+import pyloudnorm as pyln
+from typing import Optional
+from importlib import import_module
+
+from modules.encoder import (
+    StatisticReduction,
+    LogCrest,
+    LogRMS,
+    LogSpread,
+    LogSpectralBandwidth,
+    LogSpectralCentroid,
+    LogSpectralFlatness,
+    Frame,
+    MapAndMerge,
+)
+from modules.fx import hadamard
+
+
+# ------------------ Normalization  functions ------------------
+
+
+def apply_fade_in(x: torch.Tensor, num_samples: int = 16384):
+    """Apply fade in to the first num_samples of the audio signal.
+
+    Args:
+        x (torch.Tensor): Input audio tensor
+        num_samples (int, optional): Number of samples to apply fade in. Defaults to 16384.
+
+    Returns:
+        torch.Tensor: Audio tensor with fade in applied
+    """
+    fade = torch.linspace(0, 1, num_samples, device=x.device)
+    x[..., :num_samples] = x[..., :num_samples] * fade
+    return x
+
+
+def batch_peak_normalize(x: torch.Tensor):
+    peak = torch.max(torch.abs(x), dim=1)[0]
+    x = x / peak[:, None].clamp(min=1e-8)
+    return x
+
+
+def batch_loudness_normalize(x: torch.Tensor, meter: pyln.Meter, target_lufs: float):
+    for batch_idx in range(x.shape[0]):
+        lufs = meter.integrated_loudness(
+            x[batch_idx : batch_idx + 1, ...].permute(1, 0).cpu().numpy()
+        )
+        gain_db = target_lufs - lufs
+        gain_lin = 10 ** (gain_db / 20)
+        x[batch_idx, :] = gain_lin * x[batch_idx, :]
+    return x
+
+
+# -------- self-supervised parameter estimation model -------- #
+
+
+def get_param_embeds(
+    x: torch.Tensor,
+    model: torch.nn.Module,
+    sample_rate: int,
+    dropout: float = 0.0,
+):
+
+    # if peak_normalize:
+    #    x = batch_peak_normalize(x)
+
+    if sample_rate != 48000:
+        x = torchaudio.functional.resample(x, sample_rate, 48000)
+
+    seq_len = x.shape[-1]  # update seq_len after resampling
+    # if longer than 262144 crop, else repeat pad to 262144
+    # if seq_len > 262144:
+    #    x = x[:, :, :262144]
+    # else:
+    #    x = torch.nn.functional.pad(x, (0, 262144 - seq_len), "replicate")
+
+    # peak normalize each batch item
+    # for batch_idx in range(bs):
+    #     x[batch_idx, ...] /= x[batch_idx, ...].abs().max().clamp(1e-8)
+    # x = x / x.abs().amax(dim=(-1, -2), keepdim=True).clamp(min=1e-8)
+
+    mid_embeddings, side_embeddings = model(x)
+
+    # add dropout
+    if dropout > 0.0:
+        mid_embeddings = torch.nn.functional.dropout(
+            mid_embeddings, p=dropout, training=True
+        )
+        side_embeddings = torch.nn.functional.dropout(
+            side_embeddings, p=dropout, training=True
+        )
+
+    # check for nan
+    if torch.isnan(mid_embeddings).any():
+        print("Warning: NaNs found in mid_embeddings")
+        mid_embeddings = torch.nan_to_num(mid_embeddings)
+    elif torch.isnan(side_embeddings).any():
+        print("Warning: NaNs found in side_embeddings")
+        side_embeddings = torch.nan_to_num(side_embeddings)
+
+    # l2 normalize
+    mid_embeddings = torch.nn.functional.normalize(mid_embeddings, p=2, dim=-1)
+    side_embeddings = torch.nn.functional.normalize(side_embeddings, p=2, dim=-1)
+
+    return mid_embeddings, side_embeddings
+
+
+def load_param_model(ckpt_path: Optional[str] = None):
+
+    if ckpt_path is None:  # look in tmp direcory
+        ckpt_path = os.path.join(os.getcwd(), "tmp", "afx-rep.ckpt")
+        os.makedirs("tmp", exist_ok=True)
+        if not os.path.isfile(ckpt_path):
+            # download from huggingfacehub
+            os.system(
+                "wget -O tmp/afx-rep.ckpt https://huggingface.co/csteinmetz1/afx-rep/resolve/main/afx-rep.ckpt"
+            )
+            os.system(
+                "wget -O tmp/config.yaml https://huggingface.co/csteinmetz1/afx-rep/resolve/main/config.yaml"
+            )
+
+    config_path = os.path.join(os.path.dirname(ckpt_path), "config.yaml")
+
+    with open(config_path) as f:
+        config = yaml.safe_load(f)
+
+    encoder_configs = config["model"]["init_args"]["encoder"]
+
+    module_path, class_name = encoder_configs["class_path"].rsplit(".", 1)
+    module_path = module_path.replace("lcap", "st_ito")
+    module = import_module(module_path)
+    model = getattr(module, class_name)(**encoder_configs["init_args"])
+
+    checkpoint = torch.load(ckpt_path, map_location="cpu")
+
+    # load state dicts
+    state_dict = {}
+    for k, v in checkpoint["state_dict"].items():
+        if k.startswith("encoder"):
+            state_dict[k.replace("encoder.", "", 1)] = v
+
+    model.load_state_dict(state_dict)
+    model.eval()
+
+    return model
+
+
+def load_mfcc_feature_extractor():
+    transform = torch.nn.Sequential(
+        torchaudio.transforms.MFCC(
+            sample_rate=44100,
+            n_mfcc=25,
+            melkwargs={
+                "n_fft": 2048,
+                "hop_length": 1024,
+                "n_mels": 128,
+                "center": False,
+            },
+        ),
+        StatisticReduction(),
+        torch.nn.Flatten(-2, -1),
+    )
+    return transform
+
+
+def load_mir_feature_extractor():
+    transform = torch.nn.Sequential(
+        MapAndMerge(
+            [
+                torch.nn.Sequential(
+                    Frame(2048, 1024, center=False),
+                    MapAndMerge(
+                        [
+                            LogRMS(),
+                            LogCrest(),
+                            LogSpread(),
+                        ],
+                        dim=-2,
+                    ),
+                ),
+                torch.nn.Sequential(
+                    torchaudio.transforms.Spectrogram(
+                        n_fft=2048, hop_length=1024, center=False, power=1
+                    ),
+                    MapAndMerge(
+                        [
+                            LogSpectralCentroid(),
+                            LogSpectralBandwidth(),
+                            LogSpectralFlatness(),
+                        ],
+                        dim=-2,
+                    ),
+                ),
+            ],
+            dim=-2,
+        ),
+        StatisticReduction(),
+        torch.nn.Flatten(-2, -1),
+    )
+    return transform
+
+
+def get_feature_embeds(
+    x: torch.Tensor,
+    model: torch.nn.Module,
+):
+    bs, chs, seq_len = x.shape
+    assert chs == 2, "MFCC feature extractor expects stereo input"
+
+    x_ms = hadamard(x)
+
+    # Get embeddings
+    embeddings = model(x_ms)
+
+    # l2 normalize
+    embeddings = torch.nn.functional.normalize(embeddings, p=2, dim=-1)
+
+    return embeddings[:, 0], embeddings[:, 1]