Loading tests/binaural/compare_audio.pydeleted 100644 → 0 +0 −105 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 sys import warnings from typing import Tuple import numpy as np from .constants import SCRIPTS_DIR sys.path.append(str(SCRIPTS_DIR)) from pyaudio3dtools.audioarray import getdelay def compare_audio_arrays( left: np.ndarray, left_fs: int, right: np.ndarray, right_fs: int ) -> Tuple[float, float]: if left_fs != right_fs: return ValueError(f"Differing samplerates: {left_fs} vs {right_fs}!") if left.shape[1] != right.shape[1]: cmp_ch = min(left.shape[1], right.shape[1]) warnings.warn( f"Differing number of channels: {left.shape[1]} vs {right.shape[1]}! Comparing first {cmp_ch} channel(s)", category=RuntimeWarning, ) left = left[:, :cmp_ch] right = right[:, :cmp_ch] if left.shape[0] != right.shape[0]: cmp_smp = min(left.shape[0], right.shape[0]) warnings.warn( f"Warning - different durations: {left.shape[0] / left_fs:.2f}s vs {right.shape[0] / right_fs:.2f}s! Comparing first {cmp_smp / left_fs : .2f} sample(s)", category=RuntimeWarning, ) left = left[:cmp_smp, :] right = right[:cmp_smp, :] if not np.array_equal(left, right): delay = getdelay(left, right) delay_abs = np.abs(delay) # getdelay can return large values if signals are quite different # limit any delay compensation to 20 ms if delay != 0 and (delay_abs < left_fs / 50): warnings.warn( f"File B is delayed by {delay} samples ({delay*1000 / left_fs : .2f}ms)!", category=RuntimeWarning, ) # shift array left = np.roll(left, delay, axis=0) # zero shifted out samples if delay < 0: left[-np.abs(delay) :, :] = 0 elif delay > 0: left[: np.abs(delay), :] = 0 """ http://www-mmsp.ece.mcgill.ca/Documents/Software/Packages/AFsp/AFsp/CompAudio.html """ num = np.sum(left * right) den = np.sqrt(np.sum(left**2) * np.sum(right**2)) if den > 0: r = num / den else: r = np.inf snr = 10 * np.log10(1 / (1 - (r**2))) gain_b = num / np.sum(right**2) max_diff = np.abs(np.max(left - right)) else: snr = np.inf gain_b = 1 max_diff = 0 return snr, gain_b, max_diff tests/binaural/constants.py +7 −4 Original line number Diff line number Diff line Loading @@ -32,10 +32,14 @@ from pathlib import Path import re from tests.renderer.constants import ( TESTS_DIR, SCRIPTS_DIR, TEST_VECTOR_DIR, TESTV_DIR, OUTPUT_FORMATS_BINAURAL ) TESTS_DIR = Path(__file__).parent SCRIPTS_DIR = TESTS_DIR.parents[1].joinpath("scripts").resolve() TESTV_DIR = SCRIPTS_DIR.joinpath("testv") BITSTREAM_DIR = TESTS_DIR.joinpath("bitstream") DEC_ROM_DIR = TESTS_DIR.joinpath("dec_out_rom") HRTF_BINARY_DIR = SCRIPTS_DIR.joinpath("binauralRenderer_interface", "binaural_renderers_hrtf_data") Loading Loading @@ -91,7 +95,6 @@ FORMAT_TO_METADATA_FILES = { "ISM" : "stvISM{}.csv" } OUTPUT_FORMATS_BINAURAL = ["BINAURAL", "BINAURAL_ROOM_IR", "BINAURAL_ROOM_REVERB"] HR_TRAJECTORIES_TO_TEST = ["headrot_case00_3000_q", "headrot"] Loading tests/binaural/utils.py +3 −50 Original line number Diff line number Diff line Loading @@ -30,66 +30,19 @@ the United Nations Convention on Contracts on the International Sales of Goods. """ import logging import pytest import subprocess as sp import sys from typing import Dict, Optional, Tuple import numpy as np from typing import Dict, Optional import os import uuid from .constants import * from .compare_audio import compare_audio_arrays from tests.renderer.compare_audio import compare_audio_arrays from tests.renderer.utils import check_BE, test_info, run_cmd sys.path.append(SCRIPTS_DIR) import pyaudio3dtools # fixture returns test information, enabling per-testcase SNR @pytest.fixture def test_info(request): return request def run_cmd(cmd, env=None): logging.info(f"\nRunning command\n{' '.join(cmd)}\n") try: sp.run(cmd, check=True, capture_output=True, text=True) except sp.CalledProcessError as e: raise SystemError( f"Command returned non-zero exit status ({e.returncode}): {' '.join(e.cmd)}\n{e.stderr}\n{e.stdout}" ) def check_BE( test_info, afrom: np.ndarray, afrom_fs: int, afbin: np.ndarray, afbin_fs: int, ): if afrom is None or np.array_equal(afrom, np.zeros_like(afrom)): pytest.fail("Signal from ROM does not exist or is zero!") if afbin is None or np.array_equal(afbin, np.zeros_like(afbin)): pytest.fail("Signal from binary file does not exist or is zero!") snr, gain_b, max_diff = compare_audio_arrays(afrom, afrom_fs, afbin, afbin_fs) if np.isnan(snr) or gain_b == 0: pytest.fail("Invalid comparison result, check your signals!") if afrom.shape[0] < afbin.shape[0]: afrom = np.pad(afrom, [(0, afbin.shape[0] - afrom.shape[0]), (0, 0)]) elif afrom.shape[0] > afbin.shape[0]: afbin = np.pad(afbin, [(0, afrom.shape[0] - afbin.shape[0]), (0, 0)]) # check max_diff as well, since compare_audio_arrays will try to adjust for small delay differences if not np.allclose(afrom, afbin, rtol=0, atol=2) and max_diff > 2: pytest.fail( f"Signal from binary not BE to ROM! SNR : {snr:3.2f} dB, Gain : {gain_b:1.3f}, Max Diff = {int(max_diff)}" ) def run_encoder( bitrate: int, Loading tests/conftest.py +1 −1 Original line number Diff line number Diff line Loading @@ -408,7 +408,7 @@ class DecoderFrontend: if add_option_list is not None: command.extend(add_option_list) # add to fix test failed to be removed after merge # TODO add to fix test failed to be removed after merge if self._type == "REF": if '-hrtf' in command: command[command.index('-hrtf') + 1] = command[command.index('-hrtf') + 1] + "old" Loading tests/renderer/utils.py +0 −2 Original line number Diff line number Diff line Loading @@ -40,7 +40,6 @@ from typing import Dict, Optional import numpy as np import pytest from .compare_audio import compare_audio_arrays from .constants import * Loading Loading @@ -195,7 +194,6 @@ def run_renderer( if non_diegetic_pan is not None: cmd.extend(["-non_diegetic_pan", str(non_diegetic_pan)]) if refrot_file is not None: cmd.extend(["-rf", str(refrot_file)]) cmd.extend(["-otr", "ref"]) Loading Loading
tests/binaural/compare_audio.pydeleted 100644 → 0 +0 −105 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 sys import warnings from typing import Tuple import numpy as np from .constants import SCRIPTS_DIR sys.path.append(str(SCRIPTS_DIR)) from pyaudio3dtools.audioarray import getdelay def compare_audio_arrays( left: np.ndarray, left_fs: int, right: np.ndarray, right_fs: int ) -> Tuple[float, float]: if left_fs != right_fs: return ValueError(f"Differing samplerates: {left_fs} vs {right_fs}!") if left.shape[1] != right.shape[1]: cmp_ch = min(left.shape[1], right.shape[1]) warnings.warn( f"Differing number of channels: {left.shape[1]} vs {right.shape[1]}! Comparing first {cmp_ch} channel(s)", category=RuntimeWarning, ) left = left[:, :cmp_ch] right = right[:, :cmp_ch] if left.shape[0] != right.shape[0]: cmp_smp = min(left.shape[0], right.shape[0]) warnings.warn( f"Warning - different durations: {left.shape[0] / left_fs:.2f}s vs {right.shape[0] / right_fs:.2f}s! Comparing first {cmp_smp / left_fs : .2f} sample(s)", category=RuntimeWarning, ) left = left[:cmp_smp, :] right = right[:cmp_smp, :] if not np.array_equal(left, right): delay = getdelay(left, right) delay_abs = np.abs(delay) # getdelay can return large values if signals are quite different # limit any delay compensation to 20 ms if delay != 0 and (delay_abs < left_fs / 50): warnings.warn( f"File B is delayed by {delay} samples ({delay*1000 / left_fs : .2f}ms)!", category=RuntimeWarning, ) # shift array left = np.roll(left, delay, axis=0) # zero shifted out samples if delay < 0: left[-np.abs(delay) :, :] = 0 elif delay > 0: left[: np.abs(delay), :] = 0 """ http://www-mmsp.ece.mcgill.ca/Documents/Software/Packages/AFsp/AFsp/CompAudio.html """ num = np.sum(left * right) den = np.sqrt(np.sum(left**2) * np.sum(right**2)) if den > 0: r = num / den else: r = np.inf snr = 10 * np.log10(1 / (1 - (r**2))) gain_b = num / np.sum(right**2) max_diff = np.abs(np.max(left - right)) else: snr = np.inf gain_b = 1 max_diff = 0 return snr, gain_b, max_diff
tests/binaural/constants.py +7 −4 Original line number Diff line number Diff line Loading @@ -32,10 +32,14 @@ from pathlib import Path import re from tests.renderer.constants import ( TESTS_DIR, SCRIPTS_DIR, TEST_VECTOR_DIR, TESTV_DIR, OUTPUT_FORMATS_BINAURAL ) TESTS_DIR = Path(__file__).parent SCRIPTS_DIR = TESTS_DIR.parents[1].joinpath("scripts").resolve() TESTV_DIR = SCRIPTS_DIR.joinpath("testv") BITSTREAM_DIR = TESTS_DIR.joinpath("bitstream") DEC_ROM_DIR = TESTS_DIR.joinpath("dec_out_rom") HRTF_BINARY_DIR = SCRIPTS_DIR.joinpath("binauralRenderer_interface", "binaural_renderers_hrtf_data") Loading Loading @@ -91,7 +95,6 @@ FORMAT_TO_METADATA_FILES = { "ISM" : "stvISM{}.csv" } OUTPUT_FORMATS_BINAURAL = ["BINAURAL", "BINAURAL_ROOM_IR", "BINAURAL_ROOM_REVERB"] HR_TRAJECTORIES_TO_TEST = ["headrot_case00_3000_q", "headrot"] Loading
tests/binaural/utils.py +3 −50 Original line number Diff line number Diff line Loading @@ -30,66 +30,19 @@ the United Nations Convention on Contracts on the International Sales of Goods. """ import logging import pytest import subprocess as sp import sys from typing import Dict, Optional, Tuple import numpy as np from typing import Dict, Optional import os import uuid from .constants import * from .compare_audio import compare_audio_arrays from tests.renderer.compare_audio import compare_audio_arrays from tests.renderer.utils import check_BE, test_info, run_cmd sys.path.append(SCRIPTS_DIR) import pyaudio3dtools # fixture returns test information, enabling per-testcase SNR @pytest.fixture def test_info(request): return request def run_cmd(cmd, env=None): logging.info(f"\nRunning command\n{' '.join(cmd)}\n") try: sp.run(cmd, check=True, capture_output=True, text=True) except sp.CalledProcessError as e: raise SystemError( f"Command returned non-zero exit status ({e.returncode}): {' '.join(e.cmd)}\n{e.stderr}\n{e.stdout}" ) def check_BE( test_info, afrom: np.ndarray, afrom_fs: int, afbin: np.ndarray, afbin_fs: int, ): if afrom is None or np.array_equal(afrom, np.zeros_like(afrom)): pytest.fail("Signal from ROM does not exist or is zero!") if afbin is None or np.array_equal(afbin, np.zeros_like(afbin)): pytest.fail("Signal from binary file does not exist or is zero!") snr, gain_b, max_diff = compare_audio_arrays(afrom, afrom_fs, afbin, afbin_fs) if np.isnan(snr) or gain_b == 0: pytest.fail("Invalid comparison result, check your signals!") if afrom.shape[0] < afbin.shape[0]: afrom = np.pad(afrom, [(0, afbin.shape[0] - afrom.shape[0]), (0, 0)]) elif afrom.shape[0] > afbin.shape[0]: afbin = np.pad(afbin, [(0, afrom.shape[0] - afbin.shape[0]), (0, 0)]) # check max_diff as well, since compare_audio_arrays will try to adjust for small delay differences if not np.allclose(afrom, afbin, rtol=0, atol=2) and max_diff > 2: pytest.fail( f"Signal from binary not BE to ROM! SNR : {snr:3.2f} dB, Gain : {gain_b:1.3f}, Max Diff = {int(max_diff)}" ) def run_encoder( bitrate: int, Loading
tests/conftest.py +1 −1 Original line number Diff line number Diff line Loading @@ -408,7 +408,7 @@ class DecoderFrontend: if add_option_list is not None: command.extend(add_option_list) # add to fix test failed to be removed after merge # TODO add to fix test failed to be removed after merge if self._type == "REF": if '-hrtf' in command: command[command.index('-hrtf') + 1] = command[command.index('-hrtf') + 1] + "old" Loading
tests/renderer/utils.py +0 −2 Original line number Diff line number Diff line Loading @@ -40,7 +40,6 @@ from typing import Dict, Optional import numpy as np import pytest from .compare_audio import compare_audio_arrays from .constants import * Loading Loading @@ -195,7 +194,6 @@ def run_renderer( if non_diegetic_pan is not None: cmd.extend(["-non_diegetic_pan", str(non_diegetic_pan)]) if refrot_file is not None: cmd.extend(["-rf", str(refrot_file)]) cmd.extend(["-otr", "ref"]) Loading