From c4d1343897e172981350dd8bb4e224b7431fd80f Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?F=C3=A9lix=20Voituret?= Date: Fri, 8 Nov 2019 17:09:17 -0500 Subject: [PATCH] refactor: move audio package from utils to top --- setup.py | 2 +- spleeter/{utils => }/audio/__init__.py | 0 spleeter/{utils => }/audio/adapter.py | 0 spleeter/{utils => }/audio/convertor.py | 0 spleeter/{utils => }/audio/ffmpeg.py | 0 spleeter/{utils => }/audio/spectrogram.py | 0 spleeter/commands/separate.py | 4 ++-- spleeter/commands/train.py | 2 +- spleeter/dataset.py | 21 +++++++++++---------- spleeter/separator.py | 4 ++-- 10 files changed, 17 insertions(+), 16 deletions(-) rename spleeter/{utils => }/audio/__init__.py (100%) rename spleeter/{utils => }/audio/adapter.py (100%) rename spleeter/{utils => }/audio/convertor.py (100%) rename spleeter/{utils => }/audio/ffmpeg.py (100%) rename spleeter/{utils => }/audio/spectrogram.py (100%) diff --git a/setup.py b/setup.py index 6680413..b9abf78 100644 --- a/setup.py +++ b/setup.py @@ -51,13 +51,13 @@ setup( license='MIT License', packages=[ 'spleeter', + 'spleeter.audio', 'spleeter.commands', 'spleeter.model', 'spleeter.model.functions', 'spleeter.model.provider', 'spleeter.resources', 'spleeter.utils', - 'spleeter.utils.audio', ], package_data={'spleeter.resources': ['*.json']}, python_requires='>=3.6, <3.8', diff --git a/spleeter/utils/audio/__init__.py b/spleeter/audio/__init__.py similarity index 100% rename from spleeter/utils/audio/__init__.py rename to spleeter/audio/__init__.py diff --git a/spleeter/utils/audio/adapter.py b/spleeter/audio/adapter.py similarity index 100% rename from spleeter/utils/audio/adapter.py rename to spleeter/audio/adapter.py diff --git a/spleeter/utils/audio/convertor.py b/spleeter/audio/convertor.py similarity index 100% rename from spleeter/utils/audio/convertor.py rename to spleeter/audio/convertor.py diff --git a/spleeter/utils/audio/ffmpeg.py b/spleeter/audio/ffmpeg.py similarity index 100% rename from spleeter/utils/audio/ffmpeg.py rename to spleeter/audio/ffmpeg.py diff --git a/spleeter/utils/audio/spectrogram.py b/spleeter/audio/spectrogram.py similarity index 100% rename from spleeter/utils/audio/spectrogram.py rename to spleeter/audio/spectrogram.py diff --git a/spleeter/commands/separate.py b/spleeter/commands/separate.py index 71eed2d..10da43c 100644 --- a/spleeter/commands/separate.py +++ b/spleeter/commands/separate.py @@ -20,8 +20,8 @@ import tensorflow as tf import numpy as np # pylint: enable=import-error -from ..utils.audio.adapter import get_audio_adapter -from ..utils.audio.convertor import to_n_channels +from ..audio.adapter import get_audio_adapter +from ..audio.convertor import to_n_channels from ..utils.estimator import create_estimator from ..utils.tensor import set_tensor_shape diff --git a/spleeter/commands/train.py b/spleeter/commands/train.py index 2814ae6..bb48c41 100644 --- a/spleeter/commands/train.py +++ b/spleeter/commands/train.py @@ -13,9 +13,9 @@ from functools import partial import tensorflow as tf # pylint: enable=import-error +from ..audio.adapter import get_audio_adapter from ..dataset import get_training_dataset, get_validation_dataset from ..model import model_fn -from ..utils.audio.adapter import get_audio_adapter from ..utils.logging import get_logger __email__ = 'research@deezer.com' diff --git a/spleeter/dataset.py b/spleeter/dataset.py index dc65652..4f8c3e1 100644 --- a/spleeter/dataset.py +++ b/spleeter/dataset.py @@ -2,15 +2,16 @@ # coding: utf8 """ - Module for building data preprocessing pipeline using the tensorflow data - API. - Data preprocessing such as audio loading, spectrogram computation, cropping, - feature caching or data augmentation is done using a tensorflow dataset object - that output a tuple (input_, output) where: - - input_ is a dictionary with a single key that contains the (batched) mix - spectrogram of audio samples - - output is a dictionary of spectrogram of the isolated tracks (ground truth) + Module for building data preprocessing pipeline using the tensorflow + data API. Data preprocessing such as audio loading, spectrogram + computation, cropping, feature caching or data augmentation is done + using a tensorflow dataset object that output a tuple (input_, output) + where: + - input is a dictionary with a single key that contains the (batched) + mix spectrogram of audio samples + - output is a dictionary of spectrogram of the isolated tracks + (ground truth) """ import time @@ -23,10 +24,10 @@ import numpy as np import tensorflow as tf # pylint: enable=import-error -from .utils.audio.convertor import ( +from .audio.convertor import ( db_uint_spectrogram_to_gain, spectrogram_to_db_uint) -from .utils.audio.spectrogram import ( +from .audio.spectrogram import ( compute_spectrogram_tf, random_pitch_shift, random_time_stretch) diff --git a/spleeter/separator.py b/spleeter/separator.py index a238037..a5f4efd 100644 --- a/spleeter/separator.py +++ b/spleeter/separator.py @@ -20,9 +20,9 @@ from multiprocessing import Pool from pathlib import Path from os.path import join +from .audio.adapter import get_default_audio_adapter +from .audio.convertor import to_stereo from .model import model_fn -from .utils.audio.adapter import get_default_audio_adapter -from .utils.audio.convertor import to_stereo from .utils.configuration import load_configuration from .utils.estimator import create_estimator, to_predictor