diff --git a/Binaural_Loudness.ipynb b/Binaural_Loudness.ipynb
new file mode 100644
index 0000000000000000000000000000000000000000..c81c7e231ebb18ea6a13eda4dbfd130de3af5b92
--- /dev/null
+++ b/Binaural_Loudness.ipynb
@@ -0,0 +1,550 @@
+{
+ "cells": [
+ {
+ "cell_type": "code",
+ "execution_count": 1,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "# imports\n",
+ "import pandas as pd\n",
+ "import matplotlib.pyplot as plt\n",
+ "from pathlib import Path\n",
+ "import sys\n",
+ "from tempfile import TemporaryDirectory\n",
+ "from itertools import product\n",
+ "from concurrent.futures import ProcessPoolExecutor, ThreadPoolExecutor\n",
+ "import multiprocessing as mp\n",
+ "\n",
+ "IVAS_PROCESSING_SCRIPTS_DIR = \"../ivas-processing-scripts\"\n",
+ "sys.path.append(IVAS_PROCESSING_SCRIPTS_DIR)\n",
+ "\n",
+ "from ivas_processing_scripts.audiotools.wrappers.bs1770 import get_loudness\n",
+ "from ivas_processing_scripts.audiotools.audio import fromfile\n",
+ "from ivas_processing_scripts.audiotools.audiofile import read, write\n",
+ "from ivas_processing_scripts.audiotools.audioarray import resample\n",
+ "from ivas_processing_scripts.utils import run\n",
+ "\n",
+ "%matplotlib inline"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 1,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "# constants\n",
+ "IVAS_COD_CMD = [\n",
+ " \"./IVAS_cod\",\n",
+ " \"-mc\",\n",
+ " \"\", # 2 format\n",
+ " \"\", # 3 bitrate\n",
+ " \"\", # 4 fs\n",
+ " \"\", # 5 in_file\n",
+ " \"\", # 6 bs\n",
+ "]\n",
+ "IVAS_DEC_CMD = [\n",
+ " \"./IVAS_dec\",\n",
+ " \"BINAURAL\",\n",
+ " \"\", # 2 fs\n",
+ " \"\", # 3 bit\n",
+ " \"\", # 4 out_file\n",
+ "]\n",
+ "IVAS_REND_CMD = [\n",
+ " \"./IVAS_rend\",\n",
+ " \"-i\",\n",
+ " \"\", # 2 in_file\n",
+ " \"-if\",\n",
+ " \"\", # 4 in_fmt\n",
+ " \"-o\",\n",
+ " \"\", # 6 out_file\n",
+ " \"-of\",\n",
+ " \"BINAURAL\",\n",
+ "]\n",
+ "IVAS_FS = [16, 32, 48]\n",
+ "MC_MODE_TO_BITRATE_51 = {\n",
+ " \"McMASA\": [13200, 16400, 24400, 32000],\n",
+ " \"ParamMC\": [48000, 64000, 80000],\n",
+ " \"DiscMC\": [96000, 128000, 160000, 192000, 256000, 384000, 512000],\n",
+ "}\n",
+ "MC_BITRATE_TO_MODE_51 = {\n",
+ " 13200: \"McMASA\",\n",
+ " 16400: \"McMASA\",\n",
+ " 24400: \"McMASA\",\n",
+ " 32000: \"McMASA\",\n",
+ " 48000: \"ParamMC\",\n",
+ " 64000: \"ParamMC\",\n",
+ " 80000: \"ParamMC\",\n",
+ " 96000: \"DiscMC\",\n",
+ " 128000: \"DiscMC\",\n",
+ " 160000: \"DiscMC\",\n",
+ " 192000: \"DiscMC\",\n",
+ " 256000: \"DiscMC\",\n",
+ " 384000: \"DiscMC\",\n",
+ " 512000: \"DiscMC\",\n",
+ "}\n",
+ "\n",
+ "INPUT_51 = Path(\"~/ivas/items/5_1_loudness\").expanduser()"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 1,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "# helper functions\n",
+ "def process_ivas_codec(\n",
+ " format: str, bitrate: int, in_fs: int, out_fs: int, in_file: Path\n",
+ ") -> float:\n",
+ " with TemporaryDirectory() as tmp_dir:\n",
+ " tmp_dir = Path(tmp_dir)\n",
+ " tmp_in = tmp_dir.joinpath(f\"{in_file.stem}_in.wav\")\n",
+ " tmp_bs = tmp_dir.joinpath(f\"{in_file.stem}.192\")\n",
+ " tmp_out = tmp_dir.joinpath(f\"{in_file.stem}_out.wav\")\n",
+ "\n",
+ " data, fs = read(in_file)\n",
+ " if fs != in_fs * 1000:\n",
+ " data = resample(data, fs, in_fs * 1000)\n",
+ " write(tmp_in, data, in_fs * 1000)\n",
+ "\n",
+ " cod_cmd = IVAS_COD_CMD[:]\n",
+ " cod_cmd[2] = format\n",
+ " cod_cmd[3] = str(bitrate)\n",
+ " cod_cmd[4] = str(in_fs)\n",
+ " cod_cmd[5] = str(tmp_in)\n",
+ " cod_cmd[6] = str(tmp_bs)\n",
+ "\n",
+ " run(cod_cmd)\n",
+ "\n",
+ " dec_cmd = IVAS_DEC_CMD[:]\n",
+ " dec_cmd[2] = str(out_fs)\n",
+ " dec_cmd[3] = str(tmp_bs)\n",
+ " dec_cmd[4] = str(tmp_out)\n",
+ "\n",
+ " run(dec_cmd)\n",
+ "\n",
+ " audio = fromfile(\"BINAURAL\", tmp_out)\n",
+ "\n",
+ " return get_loudness(audio)\n",
+ "\n",
+ "\n",
+ "def process_ivas_rend(format: str, in_fs: int, in_file: Path):\n",
+ " with TemporaryDirectory() as tmp_dir:\n",
+ " tmp_dir = Path(tmp_dir)\n",
+ " tmp_in = tmp_dir.joinpath(f\"{in_file.stem}_in.wav\")\n",
+ " tmp_out = tmp_dir.joinpath(f\"{in_file.stem}_out.wav\")\n",
+ "\n",
+ " data, fs = read(in_file)\n",
+ " if fs != in_fs * 1000:\n",
+ " data = resample(data, fs, in_fs * 1000)\n",
+ " write(tmp_in, data, in_fs * 1000)\n",
+ "\n",
+ " cmd = IVAS_REND_CMD[:]\n",
+ " cmd[2] = str(tmp_in)\n",
+ " cmd[4] = format\n",
+ " cmd[6] = str(tmp_out)\n",
+ "\n",
+ " run(cmd)\n",
+ "\n",
+ " audio = fromfile(\"BINAURAL\", tmp_out)\n",
+ "\n",
+ " return get_loudness(audio)"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 1,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "# dataframe setup\n",
+ "DF_COLS_CODEC = [\n",
+ " \"MC Format\",\n",
+ " \"MC Mode\",\n",
+ " \"Input File\",\n",
+ " \"Bitrate\",\n",
+ " \"Sampling rate (input)\",\n",
+ " \"Sampling rate (output)\",\n",
+ " \"Loudness\",\n",
+ "]\n",
+ "DF_COLS_RENDERER = [\n",
+ " \"MC Format\",\n",
+ " \"Input File\",\n",
+ " \"Sampling rate (input)\",\n",
+ " \"Loudness\",\n",
+ "]\n",
+ "df_codec = pd.DataFrame(columns=DF_COLS_CODEC)\n",
+ "df_renderer = pd.DataFrame(columns=DF_COLS_RENDERER)"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "# Codec Evaluation\n",
+ "# If changing to ProcessPoolExecutor, macOS needs the \"fork\" context to work in a notebook\n",
+ "# In this case ThreadPoolExecutor should suffice for the mostly I/O bound workload\n",
+ "# with ProcessPoolExecutor(mp_context=mp.get_context(\"fork\")) as e:\n",
+ "with ThreadPoolExecutor() as e:\n",
+ " futures = dict()\n",
+ " for f in INPUT_51.glob(\"*.wav\"):\n",
+ " for bitrate, mc_mode in MC_BITRATE_TO_MODE_51.items():\n",
+ " for in_fs, out_fs in product(IVAS_FS, IVAS_FS):\n",
+ " # store the futures in a dict with args so we can fetch both when the task is complete\n",
+ " futures[\n",
+ " e.submit(process_ivas_codec, \"5_1\", bitrate, in_fs, out_fs, f)\n",
+ " ] = [\"5_1\", mc_mode, f.stem, bitrate, in_fs, out_fs]\n",
+ "\n",
+ " # collect results\n",
+ " for future, args in futures.items():\n",
+ " loudness, _, _ = future.result()\n",
+ "\n",
+ " record = dict(zip(DF_COLS_CODEC, [*args, loudness]))\n",
+ "\n",
+ " df_codec = pd.concat(\n",
+ " [df_codec, pd.DataFrame.from_records([record])], ignore_index=True\n",
+ " )"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "df_codec.to_markdown(Path(\"./codec_loudness.md\"), index=False)\n",
+ "df_codec"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "# Renderer Evaluation\n",
+ "# If changing to ProcessPoolExecutor, macOS needs the \"fork\" context to work in a notebook\n",
+ "# In this case ThreadPoolExecutor should suffice for the mostly I/O bound workload\n",
+ "# with ProcessPoolExecutor(mp_context=mp.get_context(\"fork\")) as e:\n",
+ "with ThreadPoolExecutor() as e:\n",
+ " futures = dict()\n",
+ " for f in INPUT_51.glob(\"*.wav\"):\n",
+ " for in_fs in IVAS_FS:\n",
+ " futures[e.submit(process_ivas_rend, \"5_1\", in_fs, f)] = [\n",
+ " \"5_1\",\n",
+ " f.stem,\n",
+ " in_fs,\n",
+ " ]\n",
+ "\n",
+ " # collect results\n",
+ " for future, args in futures.items():\n",
+ " loudness, _, _ = future.result()\n",
+ "\n",
+ " record = dict(zip(DF_COLS_RENDERER, [*args, loudness]))\n",
+ "\n",
+ " df_renderer = pd.concat(\n",
+ " [df_renderer, pd.DataFrame.from_records([record])], ignore_index=True\n",
+ " )"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": null,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "df_renderer.to_markdown(Path(\"./renderer_loudness.md\"), index=False)\n",
+ "df_renderer"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 1,
+ "metadata": {},
+ "outputs": [
+ {
+ "data": {
+ "text/html": [
+ "\n"
+ ],
+ "text/plain": []
+ },
+ "metadata": {},
+ "output_type": "display_data"
+ },
+ {
+ "data": {
+ "text/html": [
+ "\n",
+ "\n",
+ "
\n",
+ " \n",
+ " \n",
+ " | \n",
+ " MC Format | \n",
+ " MC Mode | \n",
+ " Bitrate | \n",
+ " Sampling rate (input) | \n",
+ " Sampling rate (output) | \n",
+ " Loudness | \n",
+ " Loudness (IVAS_rend) | \n",
+ " Loudness Difference (-26 LKFS) | \n",
+ " Loudness Difference (IVAS_rend) | \n",
+ "
\n",
+ " \n",
+ " \n",
+ " \n",
+ " | 0 | \n",
+ " 5_1 | \n",
+ " DiscMC | \n",
+ " 96000 | \n",
+ " 16 | \n",
+ " 16 | \n",
+ " -21.760427 | \n",
+ " -21.579920 | \n",
+ " 4.239573 | \n",
+ " -0.180507 | \n",
+ "
\n",
+ " \n",
+ " | 1 | \n",
+ " 5_1 | \n",
+ " DiscMC | \n",
+ " 96000 | \n",
+ " 16 | \n",
+ " 32 | \n",
+ " -21.885873 | \n",
+ " -21.579920 | \n",
+ " 4.114127 | \n",
+ " -0.305953 | \n",
+ "
\n",
+ " \n",
+ " | 2 | \n",
+ " 5_1 | \n",
+ " DiscMC | \n",
+ " 96000 | \n",
+ " 16 | \n",
+ " 48 | \n",
+ " -22.188446 | \n",
+ " -21.579920 | \n",
+ " 3.811554 | \n",
+ " -0.608525 | \n",
+ "
\n",
+ " \n",
+ " | 3 | \n",
+ " 5_1 | \n",
+ " DiscMC | \n",
+ " 96000 | \n",
+ " 32 | \n",
+ " 16 | \n",
+ " -21.726489 | \n",
+ " -21.420983 | \n",
+ " 4.273511 | \n",
+ " -0.305506 | \n",
+ "
\n",
+ " \n",
+ " | 4 | \n",
+ " 5_1 | \n",
+ " DiscMC | \n",
+ " 96000 | \n",
+ " 32 | \n",
+ " 32 | \n",
+ " -21.584996 | \n",
+ " -21.420983 | \n",
+ " 4.415004 | \n",
+ " -0.164013 | \n",
+ "
\n",
+ " \n",
+ " | ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ " ... | \n",
+ "
\n",
+ " \n",
+ " | 121 | \n",
+ " 5_1 | \n",
+ " ParamMC | \n",
+ " 80000 | \n",
+ " 32 | \n",
+ " 32 | \n",
+ " -20.672214 | \n",
+ " -21.420983 | \n",
+ " 5.327786 | \n",
+ " 0.748769 | \n",
+ "
\n",
+ " \n",
+ " | 122 | \n",
+ " 5_1 | \n",
+ " ParamMC | \n",
+ " 80000 | \n",
+ " 32 | \n",
+ " 48 | \n",
+ " -20.659830 | \n",
+ " -21.420983 | \n",
+ " 5.340170 | \n",
+ " 0.761153 | \n",
+ "
\n",
+ " \n",
+ " | 123 | \n",
+ " 5_1 | \n",
+ " ParamMC | \n",
+ " 80000 | \n",
+ " 48 | \n",
+ " 16 | \n",
+ " -21.071618 | \n",
+ " -21.485968 | \n",
+ " 4.928382 | \n",
+ " 0.414350 | \n",
+ "
\n",
+ " \n",
+ " | 124 | \n",
+ " 5_1 | \n",
+ " ParamMC | \n",
+ " 80000 | \n",
+ " 48 | \n",
+ " 32 | \n",
+ " -20.602037 | \n",
+ " -21.485968 | \n",
+ " 5.397963 | \n",
+ " 0.883931 | \n",
+ "
\n",
+ " \n",
+ " | 125 | \n",
+ " 5_1 | \n",
+ " ParamMC | \n",
+ " 80000 | \n",
+ " 48 | \n",
+ " 48 | \n",
+ " -20.526470 | \n",
+ " -21.485968 | \n",
+ " 5.473530 | \n",
+ " 0.959498 | \n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
126 rows × 9 columns
\n",
+ "