new notebooks to test the new features of maelzel.snd.generate, maelz…

…el.snd.filters, and maelzel.snd.vamptools

notebooks/test/test-vamptools-pyin.ipynb

@@ -0,0 +1,320 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "id": "3d2406e2-24db-4456-930d-8c9db8735880",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from maelzel.snd import vamptools\n",
+    "from maelzel.snd import audiosample\n",
+    "from maelzel.snd import deverb\n",
+    "import pitchtools as pt\n",
+    "import matplotlib.pyplot as plt\n",
+    "import sndfileio\n",
+    "from math import nan"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "5119d97c-4541-44b2-ab3e-d5d6da3f8d8a",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "samples, sr = sndfileio.sndread(\"../snd/piano-pitchtest-Cmaj.flac\")\n",
+    "samples2 = deverb.removeSustain(samples, sr)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "ac2997b7-7b13-4602-bbc1-cd0056d12b16",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Duplicate times: t0=9.827845805, t1=9.827845805\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "mnOut size: 1789\n",
+      "m_pitchTrack size: 1789\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": [
+       "PyinResult(voicedProbabilityCurve=Linear[0.023219955:10.402539683], smoothPitchCurve=Linear[0.029024944:10.402539683], f0curve=Linear[0.023219955:10.396734694], f0candidates=[(0.023s, [525.3, 261.0, 174.2, 131.6, …], [0.1242, 0.0119, 0.0008, 0.0010, …]), (0.029s, [524.5, 261.5, 174.0, 131.3, …], [0.1391, 0.0118, 0.0027, 0.0023, …]), (0.035s, [528.6, 263.7, 175.8, 131.9, …], [0.1468, 0.0080, 0.0021, 0.0025, …]), (0.041s, [526.0, 262.4, 175.4, 131.5, …], [0.2193, 0.0182, 0.0055, 0.0047, …]), (0.046s, [526.8, 263.3, 175.0, 131.2, …], [0.2278, 0.0290, 0.0149, 0.0102, …]), (0.052s, [528.3, 263.6, 175.7, 131.5, …], [0.1444, 0.0139, 0.0088, 0.0088, …]), (0.058s, [527.7, 263.5, 175.5, 131.5, …], [0.1061, 0.0092, 0.0092, 0.0101, …]), (0.064s, [528.6, 264.2, 175.8, 131.7, …], [0.1022, 0.0087, 0.0087, 0.0101, …]), (0.070s, [529.9, 264.2, 176.0, 131.9, …], [0.0947, 0.0065, 0.0076, 0.0092, …]), (0.075s, [529.0, 264.0, 175.8, 131.7, …], [0.0945, 0.0688, 0.0688, 0.1007, …]), ...]"
+      ]
+     },
+     "execution_count": 3,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "r = vamptools.pyin(samples2, sr, overlap=8, voicedThresholdPercentile=0.1, onsetSensitivity=0.1, threshDistr='beta15', lowAmpSuppressionDb=-70, preciseTime=True, maxRelativeSkew=0.1, rmsPeriod=0.020, minRmsPercentile=0.01)\n",
+    "r"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "id": "cf54baa8-9aa6-4167-99d3-541f3afe79da",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "QSocketNotifier: Can only be used with threads started with QThread\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": [
+       "<Axes: >"
+      ]
+     },
+     "execution_count": 4,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "%matplotlib qt \n",
+    "import matplotlib.pyplot as plt\n",
+    "axes = audiosample.Sample(samples, sr).plotSpectrogram(maxfreq=6000)\n",
+    "#(r.voicedProbabilityCurve * 6000).plot(axes=axes)\n",
+    "# r.f0curve.plot(axes=axes, linewidth=2, color='#ffff00', show=False)\n",
+    "c = r.smoothPitchCurveNan * (r.numCandidates > 0).nanmask()\n",
+    "c.plot(axes=axes, show=False, n=10000)\n",
+    "# r.smoothPitchCurveNan.plot(axes=axes, linewidth=2, color='#00ffff', show=False, n=10000)\n",
+    "# (r.smoothPitchCurve*2).plot(axes=axes, linewidth=2, color='#ffff00', show=False, n=1000)\n",
+    "# r.f0curve.plot(axes=axes, show=False)\n",
+    "# (r.voicedProbabilityCurve*10000).plot(axes=axes, show=False, n=10000)\n",
+    "# (r.numCandidates * 1000).plot(axes=axes, show=False, n=1000)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 28,
+   "id": "09dac2d2-f11e-491e-8bf9-95e4f83b80c2",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Linear[0.023219955:1.263219955] 6\n",
+      "Linear[1.313219955:2.4232199550000004] 6\n",
+      "Linear[2.4932199550000003:4.873219955] 24\n",
+      "Linear[5.273219955:6.583219955000001] 4\n",
+      "Linear[6.623219955000001:7.123219955000001] 2\n",
+      "Linear[7.513219955:8.063219955000001] 9\n",
+      "Linear[8.313219955000001:8.723219955000001] 7\n",
+      "Linear[8.793219955:9.023219955] 4\n",
+      "Linear[9.093219955:9.333219955] 4\n",
+      "Linear[9.373219955:9.573219954999999] 4\n",
+      "Linear[9.633219955:9.853219955] 5\n",
+      "Linear[9.883219955:10.383219955] 11\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "qt.qpa.wayland: Wayland does not support QWindow::requestActivate()\n"
+     ]
+    }
+   ],
+   "source": [
+    "parts = [b.aslinear().simplify(threshold=0.1) for b in c[::0.01].split()]\n",
+    "axes = plt.subplot()\n",
+    "for part in parts:\n",
+    "    print(part, len(part.points()[0]))\n",
+    "    part.plot(axes=axes, show=False)\n",
+    "plt.show()\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "id": "36633240-bda7-4f7e-841f-751a21c6eb60",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from math import isnan\n",
+    "\n",
+    "def split_fragments(b):\n",
+    "    xs, ys = b.points()\n",
+    "    parts = []\n",
+    "    lastpart: list[tuple[float, float]] = None\n",
+    "    for x, y in zip(xs, ys):\n",
+    "        if not isnan(y):\n",
+    "            if lastpart is None:\n",
+    "                lastpart = []\n",
+    "                parts.append(lastpart)\n",
+    "            lastpart.append((x, y))\n",
+    "        else:\n",
+    "            lastpart = None\n",
+    "    cls = b.__class__\n",
+    "    bpfs = []\n",
+    "    for part in parts:\n",
+    "        xs, ys = zip(*part)\n",
+    "        if len(xs) >= 2:\n",
+    "            bpfs.append(cls(xs, ys))\n",
+    "    return bpfs\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "id": "a528cc8b-3fe4-42ac-bd8a-7f61e8c3536f",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "\n",
+      "\n",
+      "--Csound version 6.19 (double samples) Jan 13 2024\n",
+      "[commit: c53556fbbb69d14554a4a4bd91efa441a1701d70]\n",
+      "\u001b[mlibsndfile-1.2.0\n",
+      "\u001b[msr = 44100.0,\u001b[m kr = 689.062,\u001b[m ksmps = 64\n",
+      "\u001b[m0dBFS level = 1.0,\u001b[m A4 tuning = 440.0\n",
+      "\u001b[maudio buffered in 256 sample-frame blocks\n",
+      "\u001b[mreading 2048-byte blks of shorts from adc:Built-in Audio Pro: (RAW)\n",
+      "writing 512 sample blks of 64-bit floats to dac:Built-in Audio Pro:\n",
+      "SECTION 1:\n",
+      "\u001b[m"
+     ]
+    }
+   ],
+   "source": [
+    "import csoundengine as ce\n",
+    "session = ce.Session()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "id": "d697c3af-a425-4354-9a09-b091fa7a24b0",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "session.defInstr('vco', r'''\n",
+    "|kamp=0.1, kfreq=1000, iatt=0.005, irel=0.1, isust=0.7|\n",
+    "kgain = kfreq > 100 ? 1 : 0\n",
+    "kgain = lagud:k(kgain, 0.050, 0.100)\n",
+    "aenv = linsegr:a(0, iatt, 1, 0.010, isust, irel, 0)\n",
+    "outch 1, vco2:a(kamp, kfreq) * interp(kgain) * aenv\n",
+    "''')\n",
+    "\n",
+    "def schedbpf(bpf, session):\n",
+    "    freqpairs = bpf.flatpairs()\n",
+    "    freqpairs[::2] -= bpf.x0\n",
+    "    synth = session.sched('vco', delay=bpf.x0, dur=bpf.x1 - bpf.x0, kfreq=freqpairs[1]) \n",
+    "    synth.automate('kfreq', freqpairs, delay=bpf.x0)\n",
+    "    return synth"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "id": "ea9f4b78-5943-4568-9957-2db1faeef41c",
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "qt.qpa.wayland: Wayland does not support QWindow::requestActivate()\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": [
+       "<Axes: >"
+      ]
+     },
+     "execution_count": 16,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "c = r.smoothPitchCurveNan * (r.voicedProbabilityCurve > 0.01)\n",
+    "c.plot(n=1000)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "id": "76107af5-ebd6-43c0-853c-17e1265e11fe",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import bpf4\n",
+    "parts = bpf4.util.split_fragments(r.f0curve)\n",
+    "# parts = split_fragments(r.smoothPitchCurveNan)\n",
+    "\n",
+    "with session:\n",
+    "    for part in parts:\n",
+    "        s = schedbpf(part, session=session)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "id": "116fb7bd-41b5-4bc3-a1d5-ed7ebe6fa58c",
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "[<matplotlib.lines.Line2D at 0x799ac4deb090>]"
+      ]
+     },
+     "execution_count": 11,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "prob = r.voicedProbabilityHistogram.percentileToValue(0.1)\n",
+    "axes.plot((0, r.voicedProbabilityCurve.x1), (prob*10000, prob*10000))"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "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": "3.11.6"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}