flamo 0.2.9__tar.gz → 0.2.10__tar.gz

{flamo-0.2.9 → flamo-0.2.10}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: flamo
-Version: 0.2.9
+Version: 0.2.10
 Summary: An Open-Source Library for Frequency-Domain Differentiable Audio Processing
 Project-URL: Homepage, https://github.com/gdalsanto/flamo
 Project-URL: Issues, https://github.com/gdalsanto/flamo/issues
@@ -50,7 +50,7 @@ Description-Content-Type: text/markdown
 
 Open-source library for frequency-domain differentiable audio processing.
 
-It contains differentiable implementation of common LTI audio systems modules with learnable parameters.
+It contains differentiable implementation of common LTI audio system modules with learnable parameters.
 
 ---
 
@@ -58,7 +58,7 @@ It contains differentiable implementation of common LTI audio systems modules wi
 
 Available differentiable audio signal processors - in `flamo.processor.dsp`: 
 - **Gains** : Gains, Matrices, Householder Matrices
-- **Filters** : Biquads, State Variable Filters (SVF), Graphic Equalizers (GEQ), Parametric Equiliers (PEQ - not released yet)
+- **Filters** : Biquads, State Variable Filters (SVF), Graphic Equalizers (GEQ), Parametric Equalizers (PEQ - not released yet)
 - **Delays** : Integer Delays, Fractional Delays 
 
 Transforms  - in `flamo.processor.dsp`: 
@@ -77,7 +77,7 @@ Optimization - in `flamo.optimize`:
 --- 
 
 ### 🛠️ Installation
-To install it via pip, on a new python virtual environment `flamo-env` 
+To install it via pip, on a new Python virtual environment `flamo-env` 
 ```shell
 python3.10 -m venv .flamo-env
 source .flamo-env/bin/activate
@@ -91,7 +91,7 @@ pip install flamo
 conda install -c conda-forge libsndfile
 ```
 
-For local installation: clone and install dependencies on a new pyton virtual environment `flamo-env` 
+For local installation: clone and install dependencies on a new Python virtual environment `flamo-env` 
 ```shell
 git clone https://github.com/gdalsanto/flamo
 cd flamo
@@ -160,7 +160,7 @@ model = system.Shell(core=filt, input_layer=input_layer, output_layer=output_lay
 estimation_init = model.get_freq_response()
 ````
 
-Set up optimization framework and launch it. The `Trainer` class is used to contain the model, training parameters, and training/valid steps in one class. 
+Set up the optimization framework and launch it. The `Trainer` class is used to contain the model, training parameters, and training/valid steps in one class. 
 
 ```python
 input = signal_gallery(1, n_samples=nfft, n=in_ch, signal_type='impulse', fs=fs)

{flamo-0.2.9 → flamo-0.2.10}/README.md

@@ -3,7 +3,7 @@
 
 Open-source library for frequency-domain differentiable audio processing.
 
-It contains differentiable implementation of common LTI audio systems modules with learnable parameters.
+It contains differentiable implementation of common LTI audio system modules with learnable parameters.
 
 ---
 
@@ -11,7 +11,7 @@ It contains differentiable implementation of common LTI audio systems modules wi
 
 Available differentiable audio signal processors - in `flamo.processor.dsp`: 
 - **Gains** : Gains, Matrices, Householder Matrices
-- **Filters** : Biquads, State Variable Filters (SVF), Graphic Equalizers (GEQ), Parametric Equiliers (PEQ - not released yet)
+- **Filters** : Biquads, State Variable Filters (SVF), Graphic Equalizers (GEQ), Parametric Equalizers (PEQ - not released yet)
 - **Delays** : Integer Delays, Fractional Delays 
 
 Transforms  - in `flamo.processor.dsp`: 
@@ -30,7 +30,7 @@ Optimization - in `flamo.optimize`:
 --- 
 
 ### 🛠️ Installation
-To install it via pip, on a new python virtual environment `flamo-env` 
+To install it via pip, on a new Python virtual environment `flamo-env` 
 ```shell
 python3.10 -m venv .flamo-env
 source .flamo-env/bin/activate
@@ -44,7 +44,7 @@ pip install flamo
 conda install -c conda-forge libsndfile
 ```
 
-For local installation: clone and install dependencies on a new pyton virtual environment `flamo-env` 
+For local installation: clone and install dependencies on a new Python virtual environment `flamo-env` 
 ```shell
 git clone https://github.com/gdalsanto/flamo
 cd flamo
@@ -113,7 +113,7 @@ model = system.Shell(core=filt, input_layer=input_layer, output_layer=output_lay
 estimation_init = model.get_freq_response()
 ````
 
-Set up optimization framework and launch it. The `Trainer` class is used to contain the model, training parameters, and training/valid steps in one class. 
+Set up the optimization framework and launch it. The `Trainer` class is used to contain the model, training parameters, and training/valid steps in one class. 
 
 ```python
 input = signal_gallery(1, n_samples=nfft, n=in_ch, signal_type='impulse', fs=fs)

{flamo-0.2.9 → flamo-0.2.10}/flamo/processor/dsp.py

@@ -3520,3 +3520,80 @@ class GainDelay(DSP):
         delay_samples = delay_samples.to(gain.real.dtype)
         phase = torch.einsum("fo, omn -> fmn", self.omega, delay_samples.unsqueeze(0))
         return gain.unsqueeze(0) * (self.gamma ** delay_samples) * torch.exp(-1j * phase)
+
+
+class parallelGainDelay(GainDelay):
+    """
+    Parallel counterpart of the :class:`GainDelay` class.
+    For information about **attributes** and **methods** see :class:`GainDelay`.
+
+    Shape:
+        - input: :math:`(B, M, N, ...)`
+        - param: :math:`(2, N)`
+        - output: :math:`(B, M, N, ...)`
+
+    where :math:`B` is the batch size, :math:`M` is the number of frequency bins,
+    and :math:`N` is the number of input channels.
+    Ellipsis :math:`(...)` represents additional dimensions.
+    """
+
+    def __init__(
+        self,
+        size: tuple = (1,),
+        max_len: int = 2000,
+        isint: bool = False,
+        unit: int = 100,
+        nfft: int = 2**11,
+        fs: int = 48000,
+        map_gain: Optional[callable] = None,
+        map_delay: Optional[callable] = None,
+        requires_grad: bool = False,
+        alias_decay_db: float = 0.0,
+        device: Optional[str] = None,
+        dtype: torch.dtype = torch.float32,
+    ):
+        super().__init__(
+            size=size,
+            max_len=max_len,
+            isint=isint,
+            unit=unit,
+            nfft=nfft,
+            fs=fs,
+            map_gain=map_gain,
+            map_delay=map_delay,
+            requires_grad=requires_grad,
+            alias_decay_db=alias_decay_db,
+            device=device,
+            dtype=dtype,
+        )
+
+    def check_param_shape(self):
+        """
+        Checks if the shape of the gain-delay parameters is valid.
+        """
+        assert (
+            len(self.size) == 2 and self.size[0] == 2
+        ), "parallelGainDelay parameters must have shape (2, N), for MIMO use GainDelay module."
+
+    def get_freq_convolve(self):
+        """
+        Computes the frequency convolution function for parallel gain-delay processing.
+        """
+        self.freq_convolve = lambda x, param: torch.einsum(
+            "fn,bfn...->bfn...", self.freq_response(param), x
+        )
+
+    def get_io(self):
+        """
+        Computes the number of input and output channels based on the size parameter.
+        """
+        self.input_channels = self.size[-1]
+        self.output_channels = self.size[-1]
+
+    def _combine_gain_delay(self, gain: torch.Tensor, delay_samples: torch.Tensor):
+        """
+        Combines gain and delay for parallel processing (element-wise).
+        """
+        delay_samples = delay_samples.to(gain.real.dtype)
+        phase = torch.einsum("fo, on -> fn", self.omega, delay_samples.unsqueeze(0))
+        return gain.unsqueeze(0) * (self.gamma ** delay_samples) * torch.exp(-1j * phase)

{flamo-0.2.9 → flamo-0.2.10}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "flamo"
-version = "0.2.9"
+version = "0.2.10"
 authors = [
   { name="Gloria Dal Santo", email="gloria.dalsanto@aalto.fi" },
   { name="Gian Marco De Bortoli", email="gian.debortoli@aalto.fi"},