Commit c3d0a0e9 authored by emerit's avatar emerit
Browse files

add missing files

parent 285abd86

scripts/binauralRenderer_interface/generate_crend_ivas_tables_from_sofa.c

+10 −10
Original line number Diff line number Diff line
@@ -1282,7 +1282,7 @@ int generate_crend_ivas_tables_from_sofa( const char *file_path, ConfigReader *c 
        ivas_get_hrtf_lens( ivas_hrtf, &hrtf_data_flt, frame_len );

        if ( sample_rates[indSR] != DEFAULT_SAMPLERATE )

        {

            hrtf_data.latency_s = latency_48k_optim;

            hrtf_data_flt.latency_s = latency_48k_optim;

        }



        if ( ( hrtf_data_flt.num_iterations[0][0] > ( frame_len_ms == 5 ? NUM_ITERATIONS_TO_ALLOW_OPTIM_5_MS : NUM_ITERATIONS_TO_ALLOW_OPTIM_20_MS ) ) && ( cfgReader->optimize == 1 ) )
@@ -1310,7 +1310,7 @@ int generate_crend_ivas_tables_from_sofa( const char *file_path, ConfigReader *c 


        if ( sample_rates[indSR] == DEFAULT_SAMPLERATE )

        {

            latency_48k_optim = hrtf_data.latency_s;

            latency_48k_optim = hrtf_data_flt.latency_s;

        }



        hrtf_data.index_frequency_max_diffuse = hrtf_data_flt.index_frequency_max_diffuse;
@@ -2481,7 +2481,7 @@ void update_c_file( HRTFS_DATA *hrtf, struct ivas_layout_config lscfg, const int 
    fprintf( fp, "\n};" );



    /*   float *pOut_to_bin_diffuse_re[BINAURAL_CHANNELS];*/

    if ( hrtf->pOut_to_bin_diffuse_re[0] == NULL )

    if ( hrtf->pOut_to_bin_diffuse_re_dyn[0] == NULL )

    {

        fprintf( fp, "\nconst Word16 *%s_%s_coeff_diffuse_re_%2dkHz[BINAURAL_CHANNELS]={NULL,NULL};", DECLARATION_NAME, lscfg.name, samplerate / 1000 );

    }
@@ -2495,7 +2495,7 @@ void update_c_file( HRTFS_DATA *hrtf, struct ivas_layout_config lscfg, const int 
    }



    /*   float *pOut_to_bin_diffuse_im[BINAURAL_CHANNELS];*/

    if ( hrtf->pOut_to_bin_diffuse_im[0] == NULL )

    if ( hrtf->pOut_to_bin_diffuse_im_dyn[0] == NULL )

    {

        fprintf( fp, "\nconst Word16 *%s_%s_coeff_diffuse_im_%2dkHz[BINAURAL_CHANNELS]={NULL,NULL};", DECLARATION_NAME, lscfg.name, samplerate / 1000 );

    }
@@ -2654,7 +2654,7 @@ void update_h_file( HRTFS_DATA *hrtf, struct ivas_layout_config lscfg, const int 
    fprintf( fp, "\nextern const Word16 %s_%s_inv_diffuse_weight_%2dkHz[BINAURAL_CHANNELS][%s];", DECLARATION_NAME, lscfg.name, samplerate / 1000, lscfg.output_config_num_channel_name );



    /*   uint16_t *pIndex_frequency_max_diffuse[BINAURAL_CHANNELS];*/

    if ( hrtf->pIndex_frequency_max_diffuse[0] == NULL )

    if ( hrtf->pIndex_frequency_max_diffuse_dyn[0] == NULL )

    {

        fprintf( fp, "\nextern const Word16 *%s_%s_pIndex_frequency_max_diffuse_%2dkHz[BINAURAL_CHANNELS];", DECLARATION_NAME, lscfg.name, samplerate / 1000 );

    }
@@ -2669,7 +2669,7 @@ void update_h_file( HRTFS_DATA *hrtf, struct ivas_layout_config lscfg, const int 
    fprintf( fp, "\nextern const Word16 %s_%s_coeff_im_%2dkHz[%s][BINAURAL_CHANNELS][%u];", DECLARATION_NAME, lscfg.name, samplerate / 1000, lscfg.output_config_num_channel_name, maxTotalNumFreqSampPerIterations );



    /*   float *pOut_to_bin_diffuse_re_fx[BINAURAL_CHANNELS];*/

    if ( hrtf->pOut_to_bin_diffuse_re[0] == NULL )

    if ( hrtf->pOut_to_bin_diffuse_re_dyn[0] == NULL )

    {

        fprintf( fp, "\nextern Word16 *%s_%s_coeff_diffuse_re_%2dkHz[BINAURAL_CHANNELS];", DECLARATION_NAME, lscfg.name, samplerate / 1000 );

    }
@@ -2679,7 +2679,7 @@ void update_h_file( HRTFS_DATA *hrtf, struct ivas_layout_config lscfg, const int 
    }



    /*   float *pOut_to_bin_diffuse_im_fx[BINAURAL_CHANNELS];*/

    if ( hrtf->pOut_to_bin_diffuse_im[0] == NULL )

    if ( hrtf->pOut_to_bin_diffuse_im_dyn[0] == NULL )

    {

        fprintf( fp, "\nextern Word16 *%s_%s_coeff_diffuse_im_%2dkHz[BINAURAL_CHANNELS];", DECLARATION_NAME, lscfg.name, samplerate / 1000 );

    }
@@ -2855,7 +2855,7 @@ ivas_error make_fx_be( HRTFS_DATA_flt *hrtf_flt, HRTFS_DATA *hrtf, struct ivas_l 


    maxDiff = 0;



    if ( hrtf->latency_s > 0 )

    if ( hrtf_flt->latency_s > 0 )

    {

        hrtf->factor_Q_latency_s = (int16_t) floorf( 31.f - logf( hrtf_flt->latency_s ) / logf( 2.f ) );

    }
@@ -2871,7 +2871,7 @@ ivas_error make_fx_be( HRTFS_DATA_flt *hrtf_flt, HRTFS_DATA *hrtf, struct ivas_l 
    tmp = hrtf_flt->latency_s;

    hrtf->latency_s = float2int32_t( hrtf_flt->latency_s, hrtf->factor_Q_latency_s );

    hrtf_flt->latency_s = (float) hrtf->latency_s * powf( 2.f, -1.f * (float) hrtf->factor_Q_latency_s );

    maxDiff = fabsf( hrtf->latency_s - tmp );

    maxDiff = fabsf( hrtf_flt->latency_s - tmp );



    maxVal = 0;

    for ( iIR = 0; iIR < hrtf_flt->max_num_ir; iIR++ )
@@ -3341,7 +3341,7 @@ ivas_error write_binary_file( HRTFS_DATA *hrtf, struct ivas_layout_config lscfg, 
}



/*---------------------------------------------------------------------*

 * make_fx_be();

 * make_reverb_fx_be();

 *---------------------------------------------------------------------*/

ivas_error make_reverb_fx_be( float *pEner_l, float *pEner_r, float *pCoherence, const int32_t samplerate, int16_t *factorQ )

{

scripts/binauralRenderer_interface/ivas_crend_binaural_filter_design.h

+1 −0
Original line number Diff line number Diff line
@@ -42,6 +42,7 @@ 
#define IVAS_SOFA_MAX_VAL_I      ( 1 )

#define IVAS_MAX_HRTF_LATENCY_MS ( 0.1 ) /* max hrtf latency */



#define GAIN_LFE 1.88364911f /* Gain applied to LFE during renderering */



typedef struct ivas_hrtfs_structure_flt

{

tests/cmp_pcm.py

100755 → 100644
+92 −17
Original line number Diff line number Diff line
@@ -3,6 +3,11 @@ 
import argparse

import os

import sys

import tempfile

import re

import subprocess

from pathlib import Path

from typing import Optional



THIS_PATH = os.path.join(os.getcwd(), __file__)

sys.path.append(os.path.join(os.path.dirname(THIS_PATH), "../scripts"))
@@ -10,9 +15,21 @@ sys.path.append(os.path.join(os.path.dirname(THIS_PATH), "../scripts")) 
import numpy as np

import pyaudio3dtools

import pyivastest





def cmp_pcm(file1, file2, out_config, fs, get_mld=False, mld_lim=0, abs_tol=0) -> (int, str):

from .constants import ODG_PATTERN_PQEVALAUDIO





def cmp_pcm(

    ref_file,

    cmp_file,

    out_config,

    fs,

    get_mld=False,

    allow_differing_lengths=False,

    mld_lim=0,

    abs_tol=0,

    get_ssnr=False,

    get_odg=False,

) -> (int, str):

    """

    Compare 2 PCM files for bitexactness

    """
@@ -20,36 +37,52 @@ def cmp_pcm(file1, file2, out_config, fs, get_mld=False, mld_lim=0, abs_tol=0) - 
    print("=====================")



    out_config = "MONO" if out_config == "" else out_config

    if out_config.upper() not in pyivastest.constants.OC_TO_NCHANNELS:

        out_config_in_file_names = os.path.splitext(os.path.basename(out_config))[0]

    # out_config may be a string or a Path. Wrap in str() to avoid error in case it is a Path.

    if str(out_config).upper() not in pyivastest.constants.OC_TO_NCHANNELS:

        nchannels = (

            pyivastest.IvasScriptsCommon.IvasScript.get_n_channels_from_ls_layout(

                out_config

            )

        )

    else:

        out_config_in_file_names = out_config

        nchannels = pyivastest.constants.OC_TO_NCHANNELS[out_config.upper()]



    s1, _ = pyaudio3dtools.audiofile.readfile(file1, nchannels, fs, outdtype=np.int16)

    s2, _ = pyaudio3dtools.audiofile.readfile(file2, nchannels, fs, outdtype=np.int16)

    s1, _ = pyaudio3dtools.audiofile.readfile(

        ref_file, nchannels, fs, outdtype=np.int16

    )

    s2, _ = pyaudio3dtools.audiofile.readfile(

        cmp_file, nchannels, fs, outdtype=np.int16

    )



    nchannels = s1.shape[

        1

    ]  # In case of wav input, override the nchannels with the one from the wav header

    # In case of wav input, override the nchannels with the one from the wav header

    nchannels = s1.shape[1]



    if s1.shape != s2.shape:

    if allow_differing_lengths:

        # to allow for MLD comparison, shorten longer file

        min_len = min(s1.shape[0], s2.shape[0])

        s1 = s1[:min_len, :]

        s2 = s2[:min_len, :]

    elif s1.shape != s2.shape:

        print(

            f"file size in samples: file 1 = {s1.shape[0]},",

            f"file 2 = {s2.shape[0]}",

        )

        return 1, "FAIL: File lengths differ"

        reason = "FAIL: File lengths differ. MAXIMUM ABS DIFF: None"

        if get_mld:

            reason += " - MLD: None"



        return 1, reason



    cmp_result = pyaudio3dtools.audioarray.compare(

        s1, s2, fs, per_frame=False, get_mld=get_mld

        s1,

        s2,

        fs,

        per_frame=False,

        get_mld=get_mld,

        get_ssnr=get_ssnr,

        ssnr_thresh_low=-50,

    )





    output_differs = 0

    reason = "SUCCESS: Files are bitexact"
@@ -75,13 +108,55 @@ def cmp_pcm(file1, file2, out_config, fs, get_mld=False, mld_lim=0, abs_tol=0) - 
        else:

            reason += f" > {mld_lim}"



    if get_ssnr:

        reason += " - "

        for i, s in enumerate(cmp_result["SSNR"], start=1):

            msg = f"Channel {i} SSNR: {s}"

            reason += msg + " - "



    if get_odg:

        for n in range(nchannels):

            pqeval_output = pqevalaudio_wrapper(s1[:, n], s2[:, n], fs)



            match_odg = re.search(ODG_PATTERN_PQEVALAUDIO, pqeval_output)

            odg = float(match_odg.groups()[0])

            msg = f"Channel {n} ODG: {odg}"

            reason += " - " + msg

            print(msg)



    return output_differs, reason





def pqevalaudio_wrapper(

        ref_sig: np.ndarray,

        eval_sig: np.ndarray,

        fs: int,

        ) -> str:

    with tempfile.TemporaryDirectory() as tmp_dir:

        tmp_dir = Path(tmp_dir)

        tmp_file_ref = str(tmp_dir.joinpath("ref.wav"))

        tmp_file_eval = str(tmp_dir.joinpath("eval.wav"))



        # PQevalAudio neeeds 48 kHz sampling rate

        r48 = np.clip( pyaudio3dtools.audioarray.resample(ref_sig.astype(float), fs, 48000), -32768, 32767 ).astype(np.int16)

        t48 = np.clip( pyaudio3dtools.audioarray.resample(eval_sig.astype(float), fs, 48000), -32768, 32767 ).astype(np.int16)



        pyaudio3dtools.audiofile.writefile(tmp_file_ref, r48, 48000)

        pyaudio3dtools.audiofile.writefile(tmp_file_eval, t48, 48000)



        cmd = ["PQevalAudio", tmp_file_ref, tmp_file_eval]



        result = subprocess.run(cmd, capture_output=True)

        if result.returncode != 0:

            raise RuntimeError(f"Error running PQevalaudio: {result.stderr}")



        return result.stdout.decode("utf8")





if __name__ == "__main__":

    parser = argparse.ArgumentParser()

    parser.add_argument("file1", type=str)

    parser.add_argument("file2", type=str)

    parser.add_argument("ref_file", type=str)

    parser.add_argument("cmp_file", type=str)

    parser.add_argument(

        "-o",

        "--out_config",

tests/conftest.py

+157 −13
Original line number Diff line number Diff line
@@ -34,6 +34,7 @@ Pytest customization (configuration and fixtures) for the IVAS codec test suite. 


import logging

import os

import re

from tests import testconfig

import pytest

import platform
@@ -41,6 +42,7 @@ import textwrap 
from pathlib import Path

from subprocess import TimeoutExpired, run

from typing import Optional, Union

from .constants import MLD_PATTERN, MAX_DIFF_PATTERN, SSNR_PATTERN, ODG_PATTERN



logger = logging.getLogger(__name__)

USE_LOGGER_FOR_DBG = False  # current tests do not make use of the logger feature
@@ -128,6 +130,7 @@ def pytest_addoption(parser): 
    parser.addoption(

        "--param_file",

        action="store",

        type=Path,

        help="If specified, use given param file in test_param_file.",

    )
@@ -170,6 +173,19 @@ def pytest_addoption(parser): 
        help="MLD limit for comparison (default: 0)",

        default="0",

    )



    parser.addoption(

        "--ssnr",

        action="store_true",

        help="Compute Segmental SNR (SSNR) between ref and dut output instead of just comparing for bitexactness",

    )



    parser.addoption(

        "--odg",

        action="store_true",

        help="Get Objective Difference Grade for both conditions during comparison and report difference",

    )



    parser.addoption(

        "--create_ref",

        action="store_true",
@@ -182,6 +198,19 @@ def pytest_addoption(parser): 
        default=False,

    )



    parser.addoption(

        "--use_ltv",

        action="store_true",

        default=False,

    )



    parser.addoption(

        "--ltv_dir",

        action="store",

        type=Path,

        default=None,

    )



    parser.addoption(

        "--dut_fr",

        help="Render frame size for the DUT output.",
@@ -241,6 +270,22 @@ def get_mld_lim(request): 
    return float(request.config.getoption("--mld-lim"))





@pytest.fixture(scope="session", autouse=True)

def get_ssnr(request):

    """

    Return indication to compute ssnr during ref/dut comparison.

    """

    return request.config.option.ssnr





@pytest.fixture(scope="session", autouse=True)

def get_odg(request):

    """

    Return indication to compute ssnr during ref/dut comparison.

    """

    return request.config.option.odg





@pytest.fixture(scope="session")

def abs_tol(request) -> int:

    """
@@ -312,6 +357,7 @@ class EncoderFrontend: 
        pca: Optional[bool] = None,

        quiet_mode: Optional[bool] = True,

        add_option_list: Optional[list] = None,

        run_dir: Optional[Path] = None,

    ) -> None:

        command = [self._path]
@@ -347,7 +393,11 @@ class EncoderFrontend: 


        try:

            result = run(

                command, capture_output=True, check=False, timeout=self.timeout

                command,

                capture_output=True,

                check=False,

                timeout=self.timeout,

                cwd=run_dir,

            )

        except TimeoutExpired:

            pytest.fail(f"{self._type} encoder run timed out after {self.timeout}s.")
@@ -485,7 +535,9 @@ class DecoderFrontend: 
        output_path: Path,

        quiet_mode: Optional[bool] = True,

        plc_file: Optional[Path] = None,

        netsim_profile: Optional[Path] = None,

        add_option_list: Optional[list] = None,

        run_dir: Optional[Path] = None,

    ) -> None:

        command = [self._path]
@@ -512,7 +564,7 @@ class DecoderFrontend: 


            eid_command = [eid_path]

            eid_command.extend(["-fer", "-vbr", "-bs", "g192", "-ep", "g192"])

            eid_output_suffix = "." + os.path.basename(plc_file) + ".fer"

            eid_output_suffix = ".fer"

            eid_command += [

                str(input_bitstream_path),

                str(plc_file),
@@ -521,12 +573,53 @@ class DecoderFrontend: 


            try:

                if not os.path.exists(str(input_bitstream_path) + eid_output_suffix):

                    result = run(eid_command, check=True)

            except Exception:

                pytest.fail("eid-xor operation failed!")

                    result = run(eid_command, check=True, cwd=run_dir)

            except Exception as e:

                pytest.fail(f"eid-xor operation failed! - {e}")



            input_bitstream_path += eid_output_suffix



        if netsim_profile is not None:

            system = platform.system()



            # TODO: centralize this in a utils file

            if system == "Windows":

                netsim_path = "./scripts/tools/Win32/networkSimulator_g192.exe"

            elif system == "Linux":

                netsim_path = "./scripts/tools/Linux/networkSimulator_g192"

            elif system == "Darwin":

                netsim_path = "./scripts/tools/Darwin/networkSimulator_g192"

            else:

                raise ValueError(f'Wrong system "{system}"!')



            if not os.path.isfile(netsim_path):

                raise FileNotFoundError(

                    f"network simulator binary {netsim_path} not found!\n"

                )

            netsim_bitstream_path = input_bitstream_path.with_suffix(".netsimout")

            tracefile_path = input_bitstream_path.with_suffix(".netsimtrace")

            # TODO: need to check if the "1" works with every profile

            netsim_command = [

                netsim_path,

                netsim_profile,

                input_bitstream_path,

                netsim_bitstream_path,

                tracefile_path,

                "1",

            ]

            print(netsim_command)

            try:

                run(netsim_command, check=True, cwd=run_dir)

            except Exception as e:

                pytest.fail(f"netsim operation failed! - {e}")



            input_bitstream_path = netsim_bitstream_path

            voip_opt = ["-voip"]

            if add_option_list is None:

                add_option_list = voip_opt

            elif "-voip" not in add_option_list:

                add_option_list.extend(voip_opt)



        if add_option_list is not None:

            command.extend(add_option_list)
@@ -545,7 +638,11 @@ class DecoderFrontend: 


        try:

            result = run(

                command, capture_output=True, check=False, timeout=self.timeout

                command,

                capture_output=True,

                check=False,

                timeout=self.timeout,

                cwd=run_dir,

            )

        except TimeoutExpired:

            pytest.fail(f"{self._type} decoder run timed out after {self.timeout}s.")
@@ -727,15 +824,10 @@ def decoder_only(request) -> bool: 


def pytest_configure(config):

    config.addinivalue_line("markers", "serial: mark test to run only in serial")

    config.addinivalue_line(

        "markers", "create_ref: mark test capable of producing references"

    )

    config.addinivalue_line(

        "markers",

        "create_ref_part2: reference creation test that depends on create_ref references",

    )

    if config.option.param_file:

        testconfig.PARAM_FILE = config.option.param_file

    if config.option.use_ltv:

        testconfig.use_ltv = True

    if config.option.selection_be_md5_file:

        md5_file_path = config.option.selection_be_md5_file

        if not platform.system() == "Windows":
@@ -746,3 +838,55 @@ def pytest_configure(config): 
            testconfig.MD5_REF_DICT = {

                line.split()[0]: line.split()[1] for line in f.readlines()

            }





@pytest.fixture(scope="session")

def props_to_record(request, get_mld, get_ssnr, get_odg) -> str:

    props = ["MAXIMUM ABS DIFF"]

    if get_mld:

        props.append("MLD")

    if get_ssnr:

        props.append("SSNR")

    if get_odg:

        props.append("ODG")



    return props





def parse_properties(text_to_parse: str, output_differs: bool, props_to_record: list):

    """

    Record the given properties in the report by parsing their values from the text.

    """



    props = dict()



    for prop in props_to_record:

        if prop == "MLD":

            mld = float(re.search(MLD_PATTERN, text_to_parse).groups(1)[0])

            props[prop] = mld

        elif prop == "MAXIMUM ABS DIFF":

            max_diff = 0

            if output_differs:

                if (match := re.search(MAX_DIFF_PATTERN, text_to_parse)) is not None:

                    max_diff = match.groups(1)[0]

                else:

                    raise MaxDiffPatternNotFound()

            props[prop] = max_diff

        elif prop == "SSNR":

            ssnrs = re.findall(SSNR_PATTERN, text_to_parse)

            min_ssnr = min(ssnrs)

            min_ssnr_channel = ssnrs.index(min_ssnr)

            props["MIN_SSNR"] = min_ssnr

            props["MIN_SSNR_CHANNEL"] = min_ssnr_channel

        elif prop == "ODG":

            odgs = re.findall(ODG_PATTERN, text_to_parse)

            min_odg = min(odgs)

            min_odg_channel = odgs.index(min_odg)

            props["MIN_ODG"] = min_odg

            props["MIN_ODG_CHANNEL"] = min_odg_channel



    return props





class MaxDiffPatternNotFound(Exception):

    pass

tests/constants.py

0 → 100644
+12 −0
Original line number Diff line number Diff line 
import pathlib



HERE = pathlib.Path(__file__).parent.absolute()

SCRIPTS_DIR = HERE.parent.joinpath("scripts")

TESTV_DIR = SCRIPTS_DIR.joinpath("testv")



# regex patterns for parsing the output from cmp_pcm -> mainly for BASOP ci

MLD_PATTERN = r"MLD: ([\d\.]*)"

MAX_DIFF_PATTERN = r"MAXIMUM ABS DIFF: (\d*)"

ODG_PATTERN_PQEVALAUDIO = r"Objective Difference Grade: (-*\d*\.\d*)"

ODG_PATTERN = r"ODG: (-*\d*\.\d*)"

SSNR_PATTERN = r"Channel \d* SSNR: (nan|[+-]*inf|[-*\d\.]*)"