{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import json\n",
    "from tqdm import tqdm\n",
    "from suno_utils.utils.text import write_jsonl, read_jsonl"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# load audio quality metas\n",
    "quality_metas_filepath = \"/home/christian/code/christian/metadata/genius_hq_metas_quality.json\"\n",
    "with open(quality_metas_filepath, \"r\") as f:\n",
    "    quality_metas = json.load(f)\n",
    "print(\"total quality metas: \", len(quality_metas))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "quality_metas[list(quality_metas.keys())[0]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "values = [float(meta[\"audio_quality\"][\"score\"]) for meta in quality_metas.values()]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(min(values), max(values))\n",
    "plt.hist(values, bins=100)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "subsets = [\"val\", \"tr\"]\n",
    "\n",
    "for subset in subsets:\n",
    "    found_count = 0\n",
    "    base_metas_filepath = f\"/app/suno/data/diffusion_mix/vae_25hz_30s/metas_context_aligned_{subset}.jsonl\"\n",
    "    base_metas = read_jsonl(base_metas_filepath)\n",
    "    print(f\"total {subset} metas: \", len(base_metas))\n",
    "    new_metas = []\n",
    "\n",
    "    for meta in tqdm(base_metas):\n",
    "        if meta[\"id\"] in quality_metas:\n",
    "            found_count += 1\n",
    "            new_meta = meta.copy()\n",
    "            new_meta[\"audio_quality\"] = quality_metas[meta[\"id\"]][\"audio_quality\"]\n",
    "        new_metas.append(new_meta)\n",
    "\n",
    "    print(f\"found {found_count} quality metas out of {len(base_metas)}\")\n",
    "\n",
    "    out_metas_filepath = f\"/app/suno/data/diffusion_mix/vae_25hz_30s/metas_context_aligned_quality_{subset}.jsonl\"\n",
    "    write_jsonl(new_metas, out_metas_filepath)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# make histogram of the quality scores in the new metas\n",
    "values = [float(meta[\"audio_quality\"][\"score\"]) for meta in new_metas]\n",
    "print(min(values), max(values))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# put some space between bins in the histogram\n",
    "plt.hist(values, bins=[-.3, 0.0, 2.0], rwidth=0.8)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# percent of values less than 0.0\n",
    "print(sum(1 for value in values if value < 0.0) / len(values))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "suno_env",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}