Audyn 0.0.1.dev1__tar.gz → 0.0.1.dev7__tar.gz

{Audyn-0.0.1.dev1 → audyn-0.0.1.dev7}/Audyn.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: Audyn
-Version: 0.0.1.dev1
+Version: 0.0.1.dev7
 Summary: A PyTorch toolkit for audio synthesis.
 Author: Takuya Hasumi
 License:                                  Apache License
@@ -209,6 +209,7 @@ Project-URL: url, https://github.com/tky823/Audyn
 Requires-Python: <4,>=3.8
 Description-Content-Type: text/markdown
 License-File: LICENSE
+Requires-Dist: importlib_resources<6.2; python_version == "3.8"
 Requires-Dist: omegaconf
 Requires-Dist: hydra-core
 Requires-Dist: matplotlib
@@ -217,17 +218,27 @@ Requires-Dist: torchaudio
 Requires-Dist: torchtext
 Requires-Dist: torchvision
 Requires-Dist: tensorboard
+Requires-Dist: webdataset
+Requires-Dist: inflect
 Provides-Extra: recipes
 Requires-Dist: tqdm; extra == "recipes"
 Provides-Extra: dev
 Requires-Dist: flake8; extra == "dev"
 Requires-Dist: black; extra == "dev"
 Requires-Dist: isort; extra == "dev"
+Provides-Extra: docs
+Requires-Dist: sphinx; extra == "docs"
+Requires-Dist: sphinx-autodoc-typehints; extra == "docs"
+Requires-Dist: sphinx-autobuild; extra == "docs"
+Requires-Dist: nbsphinx; extra == "docs"
+Requires-Dist: furo; extra == "docs"
 Provides-Extra: tests
 Requires-Dist: pytest; extra == "tests"
 Requires-Dist: pytest-cov; extra == "tests"
 
 # Audyn
+[![codecov](https://codecov.io/gh/tky823/Audyn/graph/badge.svg?token=7R29QDGXLQ)](https://codecov.io/gh/tky823/Audyn)
+
 Audyn is PyTorch toolkit for audio synthesis.
 
 ## Installation
@@ -248,12 +259,43 @@ If you need to run recipes, add `[recipes]` as follows:
 pip install -e ".[recipes]"
 ```
 
+If you use MacOS, you may need to set `MACOSX_DEPLOYMENT_TARGET` during installation to build C++ related modules.
+
+### C++ extension
+We use [C++ extension](https://pytorch.org/tutorials/advanced/cpp_extension.html) to search monotonic alignment in some models (e.g. GlowTTS).
+To take full advantage of computational efficiency, set appropriate value of `OMP_NUM_THREADS` and `CXX` during installation:
+
+```shell
+# In Audyn/
+export CXX=<PATH/TO/CPP/COMPILER>  # e.g. /usr/bin/c++
+export OMP_NUM_THREADS=<SUITABLE/VALUE/FOR/ENVIRONMENT>
+pip install -e "."
+```
+
 ## Development
 ```shell
 # In Audyn/
 pip install -e ".[recipes,dev,tests]"
 ```
 
+## Build Documentation Locally (optional)
+To build the documentation locally, you have to include `docs` when installing `Audyn`.
+```shell
+pip install -e ".[docs]"
+```
+
+When you build the documentation, run the following command.
+```shell
+cd docs/
+make html
+```
+
+Or, you can build the documentation automatically using `sphinx-autobuild`.
+```shell
+# in Audyn/
+sphinx-autobuild docs docs/_build/html
+```
+
 ## Test
 ```shell
 pytest tests/package

audyn-0.0.1.dev7/Audyn.egg-info/SOURCES.txt

@@ -0,0 +1,193 @@
+LICENSE
+MANIFEST.in
+README.md
+pyproject.toml
+setup.py
+Audyn.egg-info/PKG-INFO
+Audyn.egg-info/SOURCES.txt
+Audyn.egg-info/dependency_links.txt
+Audyn.egg-info/requires.txt
+Audyn.egg-info/top_level.txt
+audyn/__init__.py
+audyn/criterion/__init__.py
+audyn/criterion/autoregressive.py
+audyn/criterion/base.py
+audyn/criterion/contrastive.py
+audyn/criterion/distance.py
+audyn/criterion/fastspeech.py
+audyn/criterion/flow.py
+audyn/criterion/gan.py
+audyn/criterion/glowtts.py
+audyn/criterion/hifigan.py
+audyn/criterion/lsgan.py
+audyn/criterion/melspectrogram.py
+audyn/criterion/rvqvae.py
+audyn/criterion/ssast.py
+audyn/criterion/vqvae.py
+audyn/functional/__init__.py
+audyn/functional/activation.py
+audyn/functional/clustering.py
+audyn/functional/vector_quantization.py
+audyn/metrics/__init__.py
+audyn/metrics/base.py
+audyn/metrics/crossmodal.py
+audyn/metrics/retrieval.py
+audyn/models/__init__.py
+audyn/models/ast.py
+audyn/models/encodec.py
+audyn/models/fastspeech.py
+audyn/models/gan.py
+audyn/models/glowtts.py
+audyn/models/hifigan.py
+audyn/models/lextransformer.py
+audyn/models/passt.py
+audyn/models/pixelsnail.py
+audyn/models/roformer.py
+audyn/models/rvqvae.py
+audyn/models/soundstream.py
+audyn/models/ssast.py
+audyn/models/text_to_wave.py
+audyn/models/vae.py
+audyn/models/vqvae.py
+audyn/models/waveglow.py
+audyn/models/wavenet.py
+audyn/modules/activation.py
+audyn/modules/duration_predictor.py
+audyn/modules/encodec.py
+audyn/modules/fastspeech.py
+audyn/modules/film.py
+audyn/modules/flow.py
+audyn/modules/glow.py
+audyn/modules/glowtts.py
+audyn/modules/normalization.py
+audyn/modules/passt.py
+audyn/modules/pixelcnn.py
+audyn/modules/pixelsnail.py
+audyn/modules/positional_encoding.py
+audyn/modules/rvq.py
+audyn/modules/soundstream.py
+audyn/modules/vit.py
+audyn/modules/vq.py
+audyn/modules/vqvae.py
+audyn/modules/waveglow.py
+audyn/modules/wavenet.py
+audyn/optim/lr_scheduler.py
+audyn/optim/optimizer.py
+audyn/transforms/__init__.py
+audyn/transforms/ast.py
+audyn/transforms/birdclef.py
+audyn/transforms/cqt.py
+audyn/transforms/hifigan.py
+audyn/transforms/hubert.py
+audyn/transforms/kaldi.py
+audyn/transforms/librosa.py
+audyn/utils/__init__.py
+audyn/utils/clip_grad.py
+audyn/utils/alignment/__init__.py
+audyn/utils/alignment/monotonic_align.py
+audyn/utils/data/__init__.py
+audyn/utils/data/collator.py
+audyn/utils/data/composer.py
+audyn/utils/data/dataloader.py
+audyn/utils/data/dataset.py
+audyn/utils/data/distributed.py
+audyn/utils/data/postprocess.py
+audyn/utils/data/sampler.py
+audyn/utils/data/webdataset.py
+audyn/utils/data/audioset/__init__.py
+audyn/utils/data/audioset/_download.py
+audyn/utils/data/audioset/ast.py
+audyn/utils/data/audioset/composer.py
+audyn/utils/data/audioset/dataset.py
+audyn/utils/data/audioset/distributed.py
+audyn/utils/data/audioset/sampler.py
+audyn/utils/data/birdclef/_common/__init__.py
+audyn/utils/data/birdclef/birdclef2022/__init__.py
+audyn/utils/data/birdclef/birdclef2022/_download.py
+audyn/utils/data/birdclef/birdclef2024/__init__.py
+audyn/utils/data/birdclef/birdclef2024/_download.py
+audyn/utils/data/birdclef/birdclef2024/collator.py
+audyn/utils/data/birdclef/birdclef2024/composer.py
+audyn/utils/data/birdclef/birdclef2024/dataset.py
+audyn/utils/data/birdclef/birdclef2024/models/baseline.py
+audyn/utils/data/clotho/__init__.py
+audyn/utils/data/clotho/text/indexing.py
+audyn/utils/data/clotho/text/normalization.py
+audyn/utils/data/clotho/text/symbols.py
+audyn/utils/data/clotho/text/tokenization.py
+audyn/utils/data/cmudict/__init__.py
+audyn/utils/data/cmudict/indexing.py
+audyn/utils/data/tacotron/__init__.py
+audyn/utils/data/tacotron/text/indexing.py
+audyn/utils/data/tacotron/text/normalization.py
+audyn/utils/data/tacotron/text/numbers.py
+audyn/utils/data/tacotron/text/symbols.py
+audyn/utils/data/tacotron/text/tokenization.py
+audyn/utils/distributed/__init__.py
+audyn/utils/driver/__init__.py
+audyn/utils/driver/_decorator.py
+audyn/utils/driver/base.py
+audyn/utils/driver/feat_to_wave.py
+audyn/utils/driver/gan.py
+audyn/utils/driver/text_to_feat.py
+audyn/utils/driver/text_to_wave.py
+audyn/utils/driver/_conf_template/config.yaml
+audyn/utils/driver/_conf_template/criterion/cross_entropy.yaml
+audyn/utils/driver/_conf_template/criterion/vqvae.yaml
+audyn/utils/driver/_conf_template/data/defaults.yaml
+audyn/utils/driver/_conf_template/lr_scheduler/defaults.yaml
+audyn/utils/driver/_conf_template/lr_scheduler/none.yaml
+audyn/utils/driver/_conf_template/lr_scheduler/transformer.yaml
+audyn/utils/driver/_conf_template/model/defaults.yaml
+audyn/utils/driver/_conf_template/model/gan.yaml
+audyn/utils/driver/_conf_template/optimizer/adam.yaml
+audyn/utils/driver/_conf_template/optimizer/adam_ema.yaml
+audyn/utils/driver/_conf_template/optimizer/adamw.yaml
+audyn/utils/driver/_conf_template/optimizer/gan.yaml
+audyn/utils/driver/_conf_template/preprocess/defaults.yaml
+audyn/utils/driver/_conf_template/preprocess/ljspeech_text-to-feat.yaml
+audyn/utils/driver/_conf_template/preprocess/text-to-feat.yaml
+audyn/utils/driver/_conf_template/preprocess/text-to-wave.yaml
+audyn/utils/driver/_conf_template/system/cpu.yaml
+audyn/utils/driver/_conf_template/system/cuda.yaml
+audyn/utils/driver/_conf_template/system/cuda_amp.yaml
+audyn/utils/driver/_conf_template/system/cuda_ddp.yaml
+audyn/utils/driver/_conf_template/system/cuda_ddp_amp.yaml
+audyn/utils/driver/_conf_template/system/defaults.yaml
+audyn/utils/driver/_conf_template/system/mps.yaml
+audyn/utils/driver/_conf_template/test/defaults.yaml
+audyn/utils/driver/_conf_template/test/text_to_wave.yaml
+audyn/utils/driver/_conf_template/test/dataloader/defaults.yaml
+audyn/utils/driver/_conf_template/test/dataset/defaults.yaml
+audyn/utils/driver/_conf_template/test/dataset/torch.yaml
+audyn/utils/driver/_conf_template/train/defaults.yaml
+audyn/utils/driver/_conf_template/train/gan.yaml
+audyn/utils/driver/_conf_template/train/template.yaml
+audyn/utils/driver/_conf_template/train/text-to-feat+pretrained_feat-to-wave.yaml
+audyn/utils/driver/_conf_template/train/clip_gradient/defaults.yaml
+audyn/utils/driver/_conf_template/train/clip_gradient/gan.yaml
+audyn/utils/driver/_conf_template/train/dataloader/defaults.yaml
+audyn/utils/driver/_conf_template/train/dataloader/sequential-batch.yaml
+audyn/utils/driver/_conf_template/train/dataset/defaults.yaml
+audyn/utils/driver/_conf_template/train/dataset/sortable-torch.yaml
+audyn/utils/driver/_conf_template/train/dataset/torch.yaml
+audyn/utils/driver/_conf_template/train/dataset/webdataset.yaml
+audyn/utils/driver/_conf_template/train/record/defaults.yaml
+audyn/utils/driver/_conf_template/train/record/template.yaml
+audyn/utils/duration/__init__.py
+audyn/utils/github/__init__.py
+audyn/utils/hydra/__init__.py
+audyn/utils/hydra/utils.py
+audyn/utils/lab/__init__.py
+audyn/utils/logging/__init__.py
+audyn/utils/model/__init__.py
+audyn/utils/music/__init__.py
+audyn/utils/parallel/__init__.py
+audyn/utils/tensorboard/__init__.py
+audyn/utils/text/__init__.py
+audyn/utils/text/indexing.py
+audyn/utils/text/normalization.py
+audyn/utils/text/pronunciation.py
+audyn/utils/text/tokenization.py
+audyn/utils/textgrid/__init__.py
+cpp_extensions/monotonic_align/monotonic_align.cpp

{Audyn-0.0.1.dev1 → audyn-0.0.1.dev7}/Audyn.egg-info/requires.txt

@@ -6,12 +6,24 @@ torchaudio
 torchtext
 torchvision
 tensorboard
+webdataset
+inflect
+
+[:python_version == "3.8"]
+importlib_resources<6.2
 
 [dev]
 flake8
 black
 isort
 
+[docs]
+sphinx
+sphinx-autodoc-typehints
+sphinx-autobuild
+nbsphinx
+furo
+
 [recipes]
 tqdm
 

{Audyn-0.0.1.dev1 → audyn-0.0.1.dev7}/MANIFEST.in

@@ -1,6 +1,7 @@
 exclude .gitignore
-exclude *.yaml
 recursive-include audyn *.py
+recursive-include audyn *.yaml
 prune .github
+prune docs
 prune tests
 prune recipes

{Audyn-0.0.1.dev1 → audyn-0.0.1.dev7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: Audyn
-Version: 0.0.1.dev1
+Version: 0.0.1.dev7
 Summary: A PyTorch toolkit for audio synthesis.
 Author: Takuya Hasumi
 License:                                  Apache License
@@ -209,6 +209,7 @@ Project-URL: url, https://github.com/tky823/Audyn
 Requires-Python: <4,>=3.8
 Description-Content-Type: text/markdown
 License-File: LICENSE
+Requires-Dist: importlib_resources<6.2; python_version == "3.8"
 Requires-Dist: omegaconf
 Requires-Dist: hydra-core
 Requires-Dist: matplotlib
@@ -217,17 +218,27 @@ Requires-Dist: torchaudio
 Requires-Dist: torchtext
 Requires-Dist: torchvision
 Requires-Dist: tensorboard
+Requires-Dist: webdataset
+Requires-Dist: inflect
 Provides-Extra: recipes
 Requires-Dist: tqdm; extra == "recipes"
 Provides-Extra: dev
 Requires-Dist: flake8; extra == "dev"
 Requires-Dist: black; extra == "dev"
 Requires-Dist: isort; extra == "dev"
+Provides-Extra: docs
+Requires-Dist: sphinx; extra == "docs"
+Requires-Dist: sphinx-autodoc-typehints; extra == "docs"
+Requires-Dist: sphinx-autobuild; extra == "docs"
+Requires-Dist: nbsphinx; extra == "docs"
+Requires-Dist: furo; extra == "docs"
 Provides-Extra: tests
 Requires-Dist: pytest; extra == "tests"
 Requires-Dist: pytest-cov; extra == "tests"
 
 # Audyn
+[![codecov](https://codecov.io/gh/tky823/Audyn/graph/badge.svg?token=7R29QDGXLQ)](https://codecov.io/gh/tky823/Audyn)
+
 Audyn is PyTorch toolkit for audio synthesis.
 
 ## Installation
@@ -248,12 +259,43 @@ If you need to run recipes, add `[recipes]` as follows:
 pip install -e ".[recipes]"
 ```
 
+If you use MacOS, you may need to set `MACOSX_DEPLOYMENT_TARGET` during installation to build C++ related modules.
+
+### C++ extension
+We use [C++ extension](https://pytorch.org/tutorials/advanced/cpp_extension.html) to search monotonic alignment in some models (e.g. GlowTTS).
+To take full advantage of computational efficiency, set appropriate value of `OMP_NUM_THREADS` and `CXX` during installation:
+
+```shell
+# In Audyn/
+export CXX=<PATH/TO/CPP/COMPILER>  # e.g. /usr/bin/c++
+export OMP_NUM_THREADS=<SUITABLE/VALUE/FOR/ENVIRONMENT>
+pip install -e "."
+```
+
 ## Development
 ```shell
 # In Audyn/
 pip install -e ".[recipes,dev,tests]"
 ```
 
+## Build Documentation Locally (optional)
+To build the documentation locally, you have to include `docs` when installing `Audyn`.
+```shell
+pip install -e ".[docs]"
+```
+
+When you build the documentation, run the following command.
+```shell
+cd docs/
+make html
+```
+
+Or, you can build the documentation automatically using `sphinx-autobuild`.
+```shell
+# in Audyn/
+sphinx-autobuild docs docs/_build/html
+```
+
 ## Test
 ```shell
 pytest tests/package

audyn-0.0.1.dev7/README.md

@@ -0,0 +1,64 @@
+# Audyn
+[![codecov](https://codecov.io/gh/tky823/Audyn/graph/badge.svg?token=7R29QDGXLQ)](https://codecov.io/gh/tky823/Audyn)
+
+Audyn is PyTorch toolkit for audio synthesis.
+
+## Installation
+You can install by pip.
+```shell
+pip install git+https://github.com/tky823/Audyn.git
+```
+or clone this repository.
+```shell
+git clone https://github.com/tky823/Audyn.git
+cd Audyn
+pip install -e .
+```
+
+If you need to run recipes, add `[recipes]` as follows:
+```shell
+# In Audyn/
+pip install -e ".[recipes]"
+```
+
+If you use MacOS, you may need to set `MACOSX_DEPLOYMENT_TARGET` during installation to build C++ related modules.
+
+### C++ extension
+We use [C++ extension](https://pytorch.org/tutorials/advanced/cpp_extension.html) to search monotonic alignment in some models (e.g. GlowTTS).
+To take full advantage of computational efficiency, set appropriate value of `OMP_NUM_THREADS` and `CXX` during installation:
+
+```shell
+# In Audyn/
+export CXX=<PATH/TO/CPP/COMPILER>  # e.g. /usr/bin/c++
+export OMP_NUM_THREADS=<SUITABLE/VALUE/FOR/ENVIRONMENT>
+pip install -e "."
+```
+
+## Development
+```shell
+# In Audyn/
+pip install -e ".[recipes,dev,tests]"
+```
+
+## Build Documentation Locally (optional)
+To build the documentation locally, you have to include `docs` when installing `Audyn`.
+```shell
+pip install -e ".[docs]"
+```
+
+When you build the documentation, run the following command.
+```shell
+cd docs/
+make html
+```
+
+Or, you can build the documentation automatically using `sphinx-autobuild`.
+```shell
+# in Audyn/
+sphinx-autobuild docs docs/_build/html
+```
+
+## Test
+```shell
+pytest tests/package
+```

audyn-0.0.1.dev7/audyn/__init__.py

@@ -0,0 +1,61 @@
+import importlib
+import operator
+import re
+from typing import Any
+
+from omegaconf import OmegaConf
+
+from .utils.hydra import main
+
+__all__ = ["__version__", "main"]
+
+__version__ = "0.0.1.dev7"
+
+# for resolver
+_whitespace_re = re.compile(r"\s+")
+_int_re = re.compile(r"^\d+$")
+_float_re = re.compile(r"^[0-9]*\.?[0-9]+([eE][-+]?[0-9]+)?$")
+
+
+def _constant_resolver(full_var_name: str) -> Any:
+    if "+" in full_var_name:
+        # TODO: generalize
+        # to support whitespace, e.g. ${const:audyn.utils.data.clotho.vocab_size + 1}
+        full_var_names = full_var_name.split("+")
+        resolved = _resolve(full_var_names[0])
+
+        for _full_var_name in full_var_names[1:]:
+            _full_var_name = _whitespace_re.sub("", _full_var_name)
+
+            if _int_re.match(_full_var_name):
+                _resolved = int(_full_var_name)
+            elif _float_re.match(_full_var_name):
+                _resolved = float(_full_var_name)
+            else:
+                raise ValueError(f"{_full_var_name} cannot be converted to int nor float.")
+
+            resolved = operator.add(resolved, _resolved)
+    else:
+        resolved = _resolve(full_var_name)
+
+    return resolved
+
+
+def _resolve(full_var_name: str) -> Any:
+    full_var_name = full_var_name.strip()
+    mod_name, var_name = full_var_name.rsplit(".", maxsplit=1)
+
+    try:
+        resolved = getattr(importlib.import_module(mod_name), var_name)
+    except ModuleNotFoundError:
+        # TODO: generalize
+        attr_name = var_name
+        mod_name, var_name = mod_name.rsplit(".", maxsplit=1)
+        imported_module = importlib.import_module(mod_name)
+        cls = getattr(imported_module, var_name)
+        resolved = getattr(cls, attr_name)
+
+    return resolved
+
+
+OmegaConf.register_new_resolver("const", _constant_resolver)

audyn-0.0.1.dev7/audyn/criterion/__init__.py

@@ -0,0 +1,6 @@
+from .base import BaseCriterionWrapper, MultiCriteria
+
+__all__ = [
+    "BaseCriterionWrapper",
+    "MultiCriteria",
+]

{Audyn-0.0.1.dev1 → audyn-0.0.1.dev7}/audyn/criterion/base.py

@@ -30,13 +30,13 @@ class MultiCriteria(nn.ModuleDict):
 
     Examples:
 
-        >>> import hydra
+        >>> import audyn
         >>> import torch
         >>> torch.manual_seed(0)
         >>> config = {
-        ...     "_target_": "audyn.criterion.base.MultiCriteria",
+        ...     "_target_": "audyn.criterion.MultiCriteria",
         ...     "mse": {
-        ...         "_target_": "audyn.criterion.base.BaseCriterionWrapper",
+        ...         "_target_": "audyn.criterion.BaseCriterionWrapper",
         ...         "criterion": {
         ...             "_target_": "torch.nn.MSELoss",
         ...             "reduction": "mean",
@@ -52,7 +52,7 @@ class MultiCriteria(nn.ModuleDict):
         ...         }
         ...     },
         ...     "mae": {
-        ...         "_target_": "audyn.criterion.base.BaseCriterionWrapper",
+        ...         "_target_": "audyn.criterion.BaseCriterionWrapper",
         ...         "criterion": {
         ...             "_target_": "torch.nn.L1Loss",
         ...             "reduction": "mean",
@@ -68,7 +68,7 @@ class MultiCriteria(nn.ModuleDict):
         ...         }
         ...     }
         >>> }
-        >>> criterion = hydra.utils.instantiate(config)
+        >>> criterion = audyn.utils.instantiate_criterion(config)
         >>> y = torch.randn((4,))
         >>> t_mse = torch.randn_like(y)
         >>> t_mae = torch.randn_like(y)