started adding second preprocessing (e29184a7) · Commits · IVAS Codec Public Collaboration / IVAS Processing Scripts

ivas_processing_scripts/init.py

+9 −9

Original line number	Diff line number	Diff line
		@@ -47,6 +47,7 @@ from ivas_processing_scripts.processing.processing import (
		concat_setup,
		concat_teardown,
		preprocess,
		preprocess_2,
		process_item,
		reorder_items_list,
		)
		@@ -95,8 +96,10 @@ def main(args):
		logger = logging_init(args, cfg)

		# Re-ordering items based on concatenation order
		if cfg.concatenate_input and cfg.concatenation_order is not None:
		cfg.items_list = reorder_items_list(cfg.items_list, cfg.concatenation_order)
		if hasattr(cfg, "preprocessing_2"):
		if cfg.preprocessing_2["concatenate_input"] and cfg.preprocessing_2.get("concatenation_order", None) is not None:
		cfg.items_list = reorder_items_list(cfg.items_list, cfg.preprocessing_2["concatenation_order"])

		# check for ISM metadata
		if cfg.input["fmt"].startswith("ISM"):
		metadata = check_ISM_metadata(
		@@ -121,16 +124,13 @@ def main(args):

		# run preprocessing only once
		if hasattr(cfg, "preprocessing"):
		preprocess(cfg, cfg.metadata_path, logger)
		preprocess(cfg, logger)

		# preprocessing on whole signal(s)
		if hasattr(cfg, "preprocessing_2"):
		# TODO: call function for concatenation, preamble and background noise
		pass

		if cfg.concatenate_input:
		# concatenate items if required
		concat_setup(cfg, logger)
		preprocess_2(cfg, logger)

		# run conditions
		for condition, out_dir, tmp_dir in zip(
		cfg.proc_chains, cfg.out_dirs, cfg.tmp_dirs
		):

ivas_processing_scripts/processing/chains.py

+59 −2

Original line number	Diff line number	Diff line
		@@ -37,6 +37,7 @@ from ivas_processing_scripts.processing.evs import EVS
		from ivas_processing_scripts.processing.ivas import IVAS
		from ivas_processing_scripts.processing.postprocessing import Postprocessing
		from ivas_processing_scripts.processing.preprocessing import Preprocessing
		from ivas_processing_scripts.processing.preprocessing_2 import Preprocessing2
		from ivas_processing_scripts.utils import list_audio


		@@ -46,6 +47,9 @@ def init_processing_chains(cfg: TestConfig) -> None:
		if hasattr(cfg, "preprocessing"):
		cfg.proc_chains.append(get_preprocessing(cfg))

		if hasattr(cfg, "preprocessing_2"):
		cfg.proc_chains.append(get_preprocessing_2(cfg))

		# other processing chains
		for cond_name, cond_cfg in cfg.conditions_to_generate.items():
		bitrates = cond_cfg.get("bitrates")
		@@ -119,6 +123,49 @@ def get_preprocessing(cfg: TestConfig) -> dict:
		return chain


		def get_preprocessing_2(cfg: TestConfig) -> dict:
		"""Mapping from test configuration to preprocessing 2 keyword arguments"""
		chain = {
		"name": "preprocessing_2",
		"processes": [],
		}

		pre2_cfg = cfg.preprocessing_2
		background_cfg = pre2_cfg.get("background_noise", None)
		if background_cfg:
		background = {
		"snr": background_cfg.get("snr", None),
		"background_noise_path": background_cfg.get("background_noise_path", None),
		"seed_delay": background_cfg.get("seed_delay", 0),
		"output_fmt": cfg.postprocessing["fmt"],
		}
		else:
		background = None

		# default to input values if preprocessing was not requested
		pre_cfg = getattr(cfg, "preprocessing", {})
		tmp_in_fs = pre_cfg.get("fs", cfg.input.get("fs"))
		tmp_in_fmt = pre_cfg.get("fmt", cfg.input["fmt"])

		chain["processes"].append(
		Preprocessing2(
		{
		"in_fs": tmp_in_fs,
		"in_fmt": tmp_in_fmt,
		"concatenate_input": pre2_cfg.get("concatenate_input", False),
		"concatenation_order": pre2_cfg.get("concatenation_order", None),
		"preamble": pre2_cfg.get("preamble", 0),
		"pad_noise_preamble": pre2_cfg.get("pad_noise_preamble", False),
		"background_noise": background,
		"in_hp50": pre2_cfg.get("hp50", False),
		"multiprocessing": cfg.multiprocessing,
		}
		)
		)

		return chain


		def get_processing_chain(
		condition: str, cfg: TestConfig, bitrate: Optional[int] = None
		) -> dict:
		@@ -215,6 +262,11 @@ def get_processing_chain(
		else:
		tx_cfg = None

		if hasattr(cfg, "preprocessing_2"): # TODO: not nice
		preamble = cfg.preprocessing_2.get("preamble", 0)
		else:
		preamble = 0

		chain["processes"].append(
		EVS(
		{
		@@ -228,7 +280,7 @@ def get_processing_chain(
		"dec_opts": dec_cfg.get("opts"),
		"multiprocessing": cfg.multiprocessing,
		"tx": tx_cfg,
		"preamble": cfg.preamble,
		"preamble": preamble,
		}
		)
		)
		@@ -265,6 +317,11 @@ def get_processing_chain(
		else:
		tx_cfg = None

		if hasattr(cfg, "preprocessing_2"): # TODO: not nice
		preamble = cfg.preprocessing_2.get("preamble", 0)
		else:
		preamble = 0

		chain["processes"].append(
		IVAS(
		{
		@@ -279,7 +336,7 @@ def get_processing_chain(
		"dec_opts": dec_cfg.get("opts"),
		"multiprocessing": cfg.multiprocessing,
		"tx": tx_cfg,
		"preamble": cfg.preamble,
		"preamble": preamble,
		}
		)
		)

ivas_processing_scripts/processing/preprocessing_2.py

0 → 100644

+52 −0

Original line number	Diff line number	Diff line
		#!/usr/bin/env python3

		#
		# (C) 2022-2023 IVAS codec Public Collaboration with portions copyright Dolby International AB, Ericsson AB,
		# Fraunhofer-Gesellschaft zur Foerderung der angewandten Forschung e.V., Huawei Technologies Co. LTD.,
		# Koninklijke Philips N.V., Nippon Telegraph and Telephone Corporation, Nokia Technologies Oy, Orange,
		# Panasonic Holdings Corporation, Qualcomm Technologies, Inc., VoiceAge Corporation, and other
		# contributors to this repository. All Rights Reserved.
		#
		# This software is protected by copyright law and by international treaties.
		# The IVAS codec Public Collaboration consisting of Dolby International AB, Ericsson AB,
		# Fraunhofer-Gesellschaft zur Foerderung der angewandten Forschung e.V., Huawei Technologies Co. LTD.,
		# Koninklijke Philips N.V., Nippon Telegraph and Telephone Corporation, Nokia Technologies Oy, Orange,
		# Panasonic Holdings Corporation, Qualcomm Technologies, Inc., VoiceAge Corporation, and other
		# contributors to this repository retain full ownership rights in their respective contributions in
		# the software. This notice grants no license of any kind, including but not limited to patent
		# license, nor is any license granted by implication, estoppel or otherwise.
		#
		# Contributors are required to enter into the IVAS codec Public Collaboration agreement before making
		# contributions.
		#
		# This software is provided "AS IS", without any express or implied warranties. The software is in the
		# development stage. It is intended exclusively for experts who have experience with such software and
		# solely for the purpose of inspection. All implied warranties of non-infringement, merchantability
		# and fitness for a particular purpose are hereby disclaimed and excluded.
		#
		# Any dispute, controversy or claim arising under or in relation to providing this software shall be
		# submitted to and settled by the final, binding jurisdiction of the courts of Munich, Germany in
		# accordance with the laws of the Federal Republic of Germany excluding its conflict of law rules and
		# the United Nations Convention on Contracts on the International Sales of Goods.
		#

		import logging
		from pathlib import Path

		from ivas_processing_scripts.processing.processing import Processing


		class Preprocessing2(Processing): # TODO
		def __init__(self, attrs: dict):
		super().__init__(attrs)
		self.name = "pre_2"

		def process(self, in_file: Path, out_file: Path, in_meta, logger: logging.Logger):
		logger.debug(f"Preprocessing2 configuration : {self.__dict__}")
		logger.debug(f"Preprocessing2 {in_file.absolute()} -> {out_file.absolute()}")

		# add preamble
		# TODO

		# add background noise
		# TODO

ivas_processing_scripts/processing/processing.py

+46 −2

Original line number	Diff line number	Diff line
		@@ -36,6 +36,7 @@ from itertools import repeat
		from pathlib import Path
		from shutil import copyfile
		from typing import Iterable, Union
		from warnings import warn

		from ivas_processing_scripts.audiotools.audiofile import concat, split
		from ivas_processing_scripts.audiotools.metadata import (
		@@ -176,7 +177,7 @@ def concat_teardown(cfg: TestConfig, logger: logging.Logger):
		return out_files, out_meta


		def preprocess(cfg, in_meta, logger):
		def preprocess(cfg, logger):
		preprocessing = cfg.proc_chains[0]
		chain = preprocessing["processes"]

		@@ -191,7 +192,7 @@ def preprocess(cfg, in_meta, logger):
		repeat(cfg.out_dirs[0]),
		repeat(chain),
		repeat(logger),
		in_meta,
		cfg.metadata_path,
		),
		None,
		"mp" if cfg.multiprocessing else None,
		@@ -214,6 +215,49 @@ def preprocess(cfg, in_meta, logger):
		cfg.out_dirs = cfg.out_dirs[1:]


		def preprocess_2(cfg, logger):
		preprocessing_2 = cfg.proc_chains[0]
		chain = preprocessing_2["processes"]

		logger.info(f" Generating condition: {preprocessing_2['name']}")

		# concatenate items if required
		concat_setup(cfg, logger)

		# run preprocessing 2
		apply_func_parallel(
		process_item,
		zip(
		cfg.items_list,
		repeat(cfg.tmp_dirs[0]),
		repeat(cfg.out_dirs[0]),
		repeat(chain),
		repeat(logger),
		cfg.metadata_path,
		),
		None,
		"mp" if cfg.multiprocessing else None,
		)

		# update the configuration to use preprocessing 2 outputs as new inputs
		cfg.items_list = list_audio( # TODO: add preprocessing_2 to list of audio
		cfg.out_dirs[0], absolute=False, select_list=getattr(cfg, "input_select", None)
		)
		if cfg.metadata_path[0] is not None:
		for item_idx in range(len(cfg.metadata_path)):
		for obj_idx in range(len(cfg.metadata_path[item_idx])):
		if cfg.metadata_path[item_idx][obj_idx]:
		cfg.metadata_path[item_idx][obj_idx] = cfg.out_dirs[0] / Path(
		f"{cfg.items_list[item_idx].stem}.wav.{obj_idx}.csv"
		)
		# remove already applied processing stage
		cfg.proc_chains = cfg.proc_chains[1:]
		cfg.tmp_dirs = cfg.tmp_dirs[1:]
		cfg.out_dirs = cfg.out_dirs[1:]

		return


		def process_item(
		in_file: Union[Path, str],
		tmp_dir: Union[Path, str],