"
]
},
"metadata": {
"image/png": {
"height": 798,
"width": 1495
}
},
"output_type": "display_data"
}
],
"source": [
"plot_stft(audio_samples / MAX_ABS_INT16 , \n",
" sample_rate=EXPECTED_SAMPLE_RATE, show_black_and_white=True)\n",
"# Note: conveniently, since the plot is in log scale, the pitch outputs \n",
"# also get converted to the log scale automatically by matplotlib.\n",
"plt.scatter(confident_pitch_outputs_x, confident_pitch_values_hz, c=\"r\")\n",
"\n",
"plt.show()"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "NskqpiHLxq6V"
},
"source": [
"# 音符に変換する\n",
"\n",
"ピッチ値を得たので、次はこの値を音符に変換することにしましょう！この作業はそれ自体が困難となる部分です。次の 2 つの項目を考慮する必要があります。\n",
"\n",
"1. 休符（歌声がないところ）\n",
"2. 各音符のサイズ（オフセット） "
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "KDOlm9PLTTjt"
},
"source": [
"### 1: 出力にゼロを追加して、歌声がない部分を示す"
]
},
{
"cell_type": "code",
"execution_count": 18,
"metadata": {
"execution": {
"iopub.execute_input": "2024-01-11T18:45:01.652391Z",
"iopub.status.busy": "2024-01-11T18:45:01.651821Z",
"iopub.status.idle": "2024-01-11T18:45:01.688759Z",
"shell.execute_reply": "2024-01-11T18:45:01.688098Z"
},
"id": "9uSQ3bJmTZmo"
},
"outputs": [],
"source": [
"pitch_outputs_and_rests = [\n",
" output2hz(p) if c >= 0.9 else 0\n",
" for i, p, c in zip(indices, pitch_outputs, confidence_outputs)\n",
"]"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "9fM0UwlsTt4w"
},
"source": [
"### 2: 音符のオフセットを追加する\n",
"\n",
"自由に歌う場合、メロディーには音符が表現する絶対ピッチ値に対するオフセットがある場合があります。したがって、予測を音符に変換するには、この潜在的なオフセットを修正する必要があります。次のコードはこれを計算しています。"
]
},
{
"cell_type": "code",
"execution_count": 19,
"metadata": {
"execution": {
"iopub.execute_input": "2024-01-11T18:45:01.692521Z",
"iopub.status.busy": "2024-01-11T18:45:01.691961Z",
"iopub.status.idle": "2024-01-11T18:45:01.698363Z",
"shell.execute_reply": "2024-01-11T18:45:01.697738Z"
},
"id": "fsJu-P5ksdFW"
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"offsets: [0.2851075707500712, 0.3700368844097355, 0.2861620445674262, 0.19608817682917135, 0.17851549283916768, 0.27334483073408933, -0.4475316266590852, -0.24651997073237908, -0.17965956403558891, -0.23060136331860548, -0.3782634107643901, -0.4725100625926686, -0.3457194541269999, -0.2436666886383776, -0.1818906877810207, -0.1348058943325796, -0.24551812662426897, -0.4454884661609313, -0.3126792745167535, -0.12241535707060791, -0.06614479972665066, -0.06702634735648871, -0.1744116301709866, -0.29365739389006507, -0.32520890458170726, -0.0564402572685907, 0.1470506338899895, 0.17167006002122775, 0.16529058740790248, 0.09569531546290477, -0.006323616641203955, -0.11799822075907684, -0.18835098459069144, -0.17934754504506145, -0.17215419157092526, -0.23695640070980062, -0.34594501002376177, -0.39380421205108007, -0.2528674895936689, -0.11009436621014146, -0.07118785365169344, -0.08042248799149121, -0.12799598588293293, -0.16227484329287023, -0.059321733849699854, 0.10667988763506742, 0.21044687793906292, 0.2931920586651131, -0.22329278631751492, -0.12365553720538713, -0.4571117360765271, -0.34864566459005175, -0.35947798653189267, -0.4313194192821328, -0.4818984494978622, 0.44220950977261, 0.45883109973128455, -0.47095522924010425, -0.3674495078498552, -0.3047205333287053, -0.31075979246441676, -0.4501401792341895, 0.3966096259778311, 0.4238116671269694, 0.4982695482795947, -0.45932030423227843, -0.4890504510576079, 0.3836871527260044, 0.4441304941600137, -0.38787359430138935, -0.24855711503570177, -0.20666386647764767, -0.23811575664822726, -0.2760223047310504, -0.3641714288169524, -0.41670715643708434, -0.41009460939710607, -0.3340427999073796, -0.26122959716860805, -0.2232610212141708, -0.19940660549943345, -0.22528914465252825, -0.2780899004513415, -0.2744434134537457, -0.25655119194333764, -0.33068201704567457, -0.4678933079416083, -0.4695116715008396, -0.1648153518015647, -0.24618840082233362, -0.48052218123024204, -0.3771743489677135, -0.32261613680665846, -0.25560535951201757, -0.24629741950576545, -0.14035005553309787, -0.16659348412100883, -0.2442749349648139, -0.236978201704666, -0.20882882578912643, -0.22637519492452896, -0.29836135937516417, -0.39081296219175243, -0.3909896476442043, -0.3650093676025108, -0.2642347521955202, -0.13023199393098395, -0.18214744283501716, -0.3020849113041564, -0.33754229827467697, -0.34391801162306024, -0.31454499496763333, -0.26713314546887545, -0.2910458297902778, -0.11686573876684037, -0.1673094354445226, -0.24345146729295664, -0.30852622314040445, -0.35647376789395935, -0.37154654069487236, -0.3600149954730796, -0.2667062802488047, -0.21902000440899627, -0.2484456507736823, -0.2774126668149748, -0.2941432754570741, -0.31118778272216474, -0.32662896348779213, -0.3053947554403962, -0.2160201109821145, -0.17343703730647775, -0.17792559965198507, -0.19880643679444177, -0.2725011871630372, -0.3152120758468442, -0.28217189623658356, -0.11595223738495974, 0.0541902144377957, 0.11488166735824024, -0.2559679399306063, 0.019302356106599916, -0.002236352401425279, 0.4468796487277231, 0.15514959977323883, 0.4207713650291751, 0.3854474319642307, 0.4373497234409669, -0.4695032097274563, -0.3662756739431998, -0.20354085369650932, -0.015043790774988963, -0.4185651697093675, -0.17896653874461066, -0.032896162706066434, -0.061098168330843805, -0.1953772325689087, -0.2545161090666568, -0.3363741032654488, -0.39191536320988973, -0.36531668408458984, -0.3489657612020167, -0.35455202891175475, -0.38925380362813655, 0.48781259374078445, -0.2820884378129733, -0.241939488189864, -0.24987341685836384, -0.3034880535179809, -0.2910712014014081, -0.2783103765422581, -0.30017802073304267, -0.23735882385318519, -0.15802705569807785, -0.1688725350672513, 0.00533368216211727, -0.2545762573057857, -0.28210347487274845, -0.29791494323556833, -0.3228332309300086, -0.3895802937323367, 0.4323827980583488, 0.17438820408041522, -0.12961415393892395, -0.2236296109730489, -0.04022635205333813, -0.4264043621594098, -0.0018987662965486152, -0.07466309859101727, -0.08664951487129002, -0.08169480367248383, -0.31617519541327965, -0.47420548422877573, 0.1502044753855003, 0.30507923857624064, 0.031032583278971515, -0.17852388186996393, -0.3371385477358615, -0.4178048549467732, -0.2023933346444835, -0.10605277224127718, -0.10771248771493447, -0.16037790997569346, -0.18698410763089868, -0.17355977250879562, -0.008242337244190878, -0.011401999431292609, -0.18767393274848132, -0.360175323324853, 0.011681766969516616, -0.1931417836124183]\n",
"ideal offset: -0.16889355771202758\n"
]
}
],
"source": [
"A4 = 440\n",
"C0 = A4 * pow(2, -4.75)\n",
"note_names = [\"C\", \"C#\", \"D\", \"D#\", \"E\", \"F\", \"F#\", \"G\", \"G#\", \"A\", \"A#\", \"B\"]\n",
"\n",
"def hz2offset(freq):\n",
" # This measures the quantization error for a single note.\n",
" if freq == 0: # Rests always have zero error.\n",
" return None\n",
" # Quantized note.\n",
" h = round(12 * math.log2(freq / C0))\n",
" return 12 * math.log2(freq / C0) - h\n",
"\n",
"\n",
"# The ideal offset is the mean quantization error for all the notes\n",
"# (excluding rests):\n",
"offsets = [hz2offset(p) for p in pitch_outputs_and_rests if p != 0]\n",
"print(\"offsets: \", offsets)\n",
"\n",
"ideal_offset = statistics.mean(offsets)\n",
"print(\"ideal offset: \", ideal_offset)\n"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "K17It_qT2DtE"
},
"source": [
"ヒューリスティックを使用して歌われた可能性の最も高い音符のシーケンスの予測を試みることができるようになりました。上記で計算された理想的なオフセットは 1 つの材料ではありますが、速度（8 個など、どれくらいの予測をたてるのか）と量子化を始める時間オフセットを知る必要もあります。単純にしておくために、異なる速度と時間オフセットを試して、量子化誤差を測定し、最終的に、この誤差を最小限に抑える値を使用します。"
]
},
{
"cell_type": "code",
"execution_count": 20,
"metadata": {
"execution": {
"iopub.execute_input": "2024-01-11T18:45:01.702108Z",
"iopub.status.busy": "2024-01-11T18:45:01.701620Z",
"iopub.status.idle": "2024-01-11T18:45:02.629951Z",
"shell.execute_reply": "2024-01-11T18:45:02.629217Z"
},
"id": "eMUTI4L52ZHA"
},
"outputs": [],
"source": [
"def quantize_predictions(group, ideal_offset):\n",
" # Group values are either 0, or a pitch in Hz.\n",
" non_zero_values = [v for v in group if v != 0]\n",
" zero_values_count = len(group) - len(non_zero_values)\n",
"\n",
" # Create a rest if 80% is silent, otherwise create a note.\n",
" if zero_values_count > 0.8 * len(group):\n",
" # Interpret as a rest. Count each dropped note as an error, weighted a bit\n",
" # worse than a badly sung note (which would 'cost' 0.5).\n",
" return 0.51 * len(non_zero_values), \"Rest\"\n",
" else:\n",
" # Interpret as note, estimating as mean of non-rest predictions.\n",
" h = round(\n",
" statistics.mean([\n",
" 12 * math.log2(freq / C0) - ideal_offset for freq in non_zero_values\n",
" ]))\n",
" octave = h // 12\n",
" n = h % 12\n",
" note = note_names[n] + str(octave)\n",
" # Quantization error is the total difference from the quantized note.\n",
" error = sum([\n",
" abs(12 * math.log2(freq / C0) - ideal_offset - h)\n",
" for freq in non_zero_values\n",
" ])\n",
" return error, note\n",
"\n",
"\n",
"def get_quantization_and_error(pitch_outputs_and_rests, predictions_per_eighth,\n",
" prediction_start_offset, ideal_offset):\n",
" # Apply the start offset - we can just add the offset as rests.\n",
" pitch_outputs_and_rests = [0] * prediction_start_offset + \\\n",
" pitch_outputs_and_rests\n",
" # Collect the predictions for each note (or rest).\n",
" groups = [\n",
" pitch_outputs_and_rests[i:i + predictions_per_eighth]\n",
" for i in range(0, len(pitch_outputs_and_rests), predictions_per_eighth)\n",
" ]\n",
"\n",
" quantization_error = 0\n",
"\n",
" notes_and_rests = []\n",
" for group in groups:\n",
" error, note_or_rest = quantize_predictions(group, ideal_offset)\n",
" quantization_error += error\n",
" notes_and_rests.append(note_or_rest)\n",
"\n",
" return quantization_error, notes_and_rests\n",
"\n",
"\n",
"best_error = float(\"inf\")\n",
"best_notes_and_rests = None\n",
"best_predictions_per_note = None\n",
"\n",
"for predictions_per_note in range(20, 65, 1):\n",
" for prediction_start_offset in range(predictions_per_note):\n",
"\n",
" error, notes_and_rests = get_quantization_and_error(\n",
" pitch_outputs_and_rests, predictions_per_note,\n",
" prediction_start_offset, ideal_offset)\n",
"\n",
" if error < best_error: \n",
" best_error = error\n",
" best_notes_and_rests = notes_and_rests\n",
" best_predictions_per_note = predictions_per_note\n",
"\n",
"# At this point, best_notes_and_rests contains the best quantization.\n",
"# Since we don't need to have rests at the beginning, let's remove these:\n",
"while best_notes_and_rests[0] == 'Rest':\n",
" best_notes_and_rests = best_notes_and_rests[1:]\n",
"# Also remove silence at the end.\n",
"while best_notes_and_rests[-1] == 'Rest':\n",
" best_notes_and_rests = best_notes_and_rests[:-1]"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "vMZbWA3aVqee"
},
"source": [
"では、量子化された音符を楽譜として書き出しましょう！\n",
"\n",
"これを行うには、[music21](http://web.mit.edu/music21/) と [Open Sheet Music Display](https://github.com/opensheetmusicdisplay/opensheetmusicdisplay) の 2 つのライブラリを使用します。\n",
"\n",
"**注意:** 単純に行えるように、ここではすべての音符の長さが同じ（半音）であると仮定しています。"
]
},
{
"cell_type": "code",
"execution_count": 21,
"metadata": {
"execution": {
"iopub.execute_input": "2024-01-11T18:45:02.634069Z",
"iopub.status.busy": "2024-01-11T18:45:02.633777Z",
"iopub.status.idle": "2024-01-11T18:45:02.639591Z",
"shell.execute_reply": "2024-01-11T18:45:02.639011Z"
},
"id": "yVrk_IOIzpQR"
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"bpm: 78.26086956521739\n"
]
}
],
"source": [
"# Creating the sheet music score.\n",
"sc = music21.stream.Score()\n",
"# Adjust the speed to match the actual singing.\n",
"bpm = 60 * 60 / best_predictions_per_note\n",
"print ('bpm: ', bpm)\n",
"a = music21.tempo.MetronomeMark(number=bpm)\n",
"sc.insert(0,a)\n",
"\n",
"for snote in best_notes_and_rests: \n",
" d = 'half'\n",
" if snote == 'Rest': \n",
" sc.append(music21.note.Rest(type=d))\n",
" else:\n",
" sc.append(music21.note.Note(snote, type=d))"
]
},
{
"cell_type": "code",
"execution_count": 22,
"metadata": {
"cellView": "both",
"execution": {
"iopub.execute_input": "2024-01-11T18:45:02.642744Z",
"iopub.status.busy": "2024-01-11T18:45:02.642523Z",
"iopub.status.idle": "2024-01-11T18:45:02.648311Z",
"shell.execute_reply": "2024-01-11T18:45:02.647676Z"
},
"id": "CEleCWHtG2s4"
},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"/tmpfs/tmp/ipykernel_61905/2305315633.py:3: DeprecationWarning: Importing display from IPython.core.display is deprecated since IPython 7.14, please import from IPython display\n",
" from IPython.core.display import display, HTML, Javascript\n"
]
}
],
"source": [
"#@title [Run this] Helper function to use Open Sheet Music Display (JS code) to show a music score\n",
"\n",
"from IPython.core.display import display, HTML, Javascript\n",
"import json, random\n",
"\n",
"def showScore(score):\n",
" xml = open(score.write('musicxml')).read()\n",
" showMusicXML(xml)\n",
" \n",
"def showMusicXML(xml):\n",
" DIV_ID = \"OSMD_div\"\n",
" display(HTML('loading OpenSheetMusicDisplay
'))\n",
" script = \"\"\"\n",
" var div_id = %%DIV_ID%%;\n",
" function loadOSMD() { \n",
" return new Promise(function(resolve, reject){\n",
" if (window.opensheetmusicdisplay) {\n",
" return resolve(window.opensheetmusicdisplay)\n",
" }\n",
" // OSMD script has a 'define' call which conflicts with requirejs\n",
" var _define = window.define // save the define object \n",
" window.define = undefined // now the loaded script will ignore requirejs\n",
" var s = document.createElement( 'script' );\n",
" s.setAttribute( 'src', \"https://cdn.jsdelivr.net/npm/opensheetmusicdisplay@0.7.6/build/opensheetmusicdisplay.min.js\" );\n",
" //s.setAttribute( 'src', \"/custom/opensheetmusicdisplay.js\" );\n",
" s.onload=function(){\n",
" window.define = _define\n",
" resolve(opensheetmusicdisplay);\n",
" };\n",
" document.body.appendChild( s ); // browser will try to load the new script tag\n",
" }) \n",
" }\n",
" loadOSMD().then((OSMD)=>{\n",
" window.openSheetMusicDisplay = new OSMD.OpenSheetMusicDisplay(div_id, {\n",
" drawingParameters: \"compacttight\"\n",
" });\n",
" openSheetMusicDisplay\n",
" .load(%%data%%)\n",
" .then(\n",
" function() {\n",
" openSheetMusicDisplay.render();\n",
" }\n",
" );\n",
" })\n",
" \"\"\".replace('%%DIV_ID%%',DIV_ID).replace('%%data%%',json.dumps(xml))\n",
" display(Javascript(script))\n",
" return"
]
},
{
"cell_type": "code",
"execution_count": 23,
"metadata": {
"execution": {
"iopub.execute_input": "2024-01-11T18:45:02.651694Z",
"iopub.status.busy": "2024-01-11T18:45:02.651161Z",
"iopub.status.idle": "2024-01-11T18:45:02.675354Z",
"shell.execute_reply": "2024-01-11T18:45:02.674610Z"
},
"id": "WTu4phq4WeAI"
},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/music21/musicxml/m21ToXml.py:510: MusicXMLWarning: is not well-formed; see isWellFormedNotation()\n",
" warnings.warn(f'{scOut} is not well-formed; see isWellFormedNotation()',\n"
]
},
{
"data": {
"text/html": [
"loading OpenSheetMusicDisplay
"
],
"text/plain": [
""
]
},
"metadata": {},
"output_type": "display_data"
},
{
"data": {
"application/javascript": [
"\n",
" var div_id = OSMD_div;\n",
" function loadOSMD() { \n",
" return new Promise(function(resolve, reject){\n",
" if (window.opensheetmusicdisplay) {\n",
" return resolve(window.opensheetmusicdisplay)\n",
" }\n",
" // OSMD script has a 'define' call which conflicts with requirejs\n",
" var _define = window.define // save the define object \n",
" window.define = undefined // now the loaded script will ignore requirejs\n",
" var s = document.createElement( 'script' );\n",
" s.setAttribute( 'src', \"https://cdn.jsdelivr.net/npm/opensheetmusicdisplay@0.7.6/build/opensheetmusicdisplay.min.js\" );\n",
" //s.setAttribute( 'src', \"/custom/opensheetmusicdisplay.js\" );\n",
" s.onload=function(){\n",
" window.define = _define\n",
" resolve(opensheetmusicdisplay);\n",
" };\n",
" document.body.appendChild( s ); // browser will try to load the new script tag\n",
" }) \n",
" }\n",
" loadOSMD().then((OSMD)=>{\n",
" window.openSheetMusicDisplay = new OSMD.OpenSheetMusicDisplay(div_id, {\n",
" drawingParameters: \"compacttight\"\n",
" });\n",
" openSheetMusicDisplay\n",
" .load(\"\\n\\n\\n Music21 Fragment \\n \\n Music21 \\n \\n 2024-01-11 \\n music21 v.8.3.0 \\n \\n \\n \\n \\n 7 \\n 40 \\n \\n \\n \\n \\n \\n \\n \\n \\n \\n \\n \\n 10080 \\n \\n 4 \\n 4 \\n \\n \\n F \\n 4 \\n \\n \\n \\n \\n \\n quarter \\n 78.26086956521739 \\n \\n \\n \\n \\n \\n C \\n 3 \\n \\n 20160 \\n half \\n \\n \\n \\n D \\n 3 \\n \\n 20160 \\n half \\n \\n \\n \\n \\n \\n \\n E \\n 3 \\n \\n 20160 \\n half \\n \\n \\n \\n F \\n 3 \\n \\n 20160 \\n half \\n \\n \\n \\n \\n \\n \\n G \\n 3 \\n \\n 20160 \\n half \\n \\n \\n \\n A \\n 3 \\n \\n 20160 \\n half \\n \\n \\n \\n \\n \\n \\n B \\n 3 \\n \\n 20160 \\n half \\n \\n \\n \\n C \\n 4 \\n \\n 20160 \\n half \\n \\n \\n light-heavy \\n \\n \\n \\n \")\n",
" .then(\n",
" function() {\n",
" openSheetMusicDisplay.render();\n",
" }\n",
" );\n",
" })\n",
" "
],
"text/plain": [
""
]
},
"metadata": {},
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"['C3', 'D3', 'E3', 'F3', 'G3', 'A3', 'B3', 'C4']\n"
]
}
],
"source": [
"# rendering the music score\n",
"showScore(sc)\n",
"print(best_notes_and_rests)"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "fGPXm6Z83U2g"
},
"source": [
"音符を MIDI ファイルに変換して、聴いてみましょう。\n",
"\n",
"このファイルを作成するには、前に作成したストリームを使用できます。"
]
},
{
"cell_type": "code",
"execution_count": 24,
"metadata": {
"execution": {
"iopub.execute_input": "2024-01-11T18:45:02.679463Z",
"iopub.status.busy": "2024-01-11T18:45:02.679224Z",
"iopub.status.idle": "2024-01-11T18:45:02.687376Z",
"shell.execute_reply": "2024-01-11T18:45:02.686801Z"
},
"id": "klYoWjgmPaod"
},
"outputs": [],
"source": [
"# Saving the recognized musical notes as a MIDI file\n",
"converted_audio_file_as_midi = converted_audio_file[:-4] + '.mid'\n",
"fp = sc.write('midi', fp=converted_audio_file_as_midi)"
]
},
{
"cell_type": "code",
"execution_count": 25,
"metadata": {
"execution": {
"iopub.execute_input": "2024-01-11T18:45:02.690739Z",
"iopub.status.busy": "2024-01-11T18:45:02.690156Z",
"iopub.status.idle": "2024-01-11T18:45:02.693615Z",
"shell.execute_reply": "2024-01-11T18:45:02.693041Z"
},
"id": "tz7Mj3Qx1lpR"
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"converted_audio_file.mid_midioutput.wav\n"
]
}
],
"source": [
"wav_from_created_midi = converted_audio_file_as_midi.replace(' ', '_') + \"_midioutput.wav\"\n",
"print(wav_from_created_midi)"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "ahss5EOiWDDp"
},
"source": [
"Colab で聴くには、wav に変換し直す必要があります。簡単な方法は、Timidty を使用することです。"
]
},
{
"cell_type": "code",
"execution_count": 26,
"metadata": {
"execution": {
"iopub.execute_input": "2024-01-11T18:45:02.697107Z",
"iopub.status.busy": "2024-01-11T18:45:02.696634Z",
"iopub.status.idle": "2024-01-11T18:45:03.088028Z",
"shell.execute_reply": "2024-01-11T18:45:03.086936Z"
},
"id": "XmeJ-UITV2nq"
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Playing converted_audio_file.mid\r\n",
"MIDI file: converted_audio_file.mid\r\n",
"Format: 1 Tracks: 2 Divisions: 1024\r\n",
"Track name: \r\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Playing time: ~16 seconds\r\n",
"Notes cut: 0\r\n",
"Notes lost totally: 0\r\n"
]
}
],
"source": [
"!timidity $converted_audio_file_as_midi -Ow -o $wav_from_created_midi"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "bnvwmyNj7kCC"
},
"source": [
"最後に、モデルが推論し、予測されたピッチから MIDI 経由で作成され、音符から作成された音声を聴いてみましょう！\n"
]
},
{
"cell_type": "code",
"execution_count": 27,
"metadata": {
"execution": {
"iopub.execute_input": "2024-01-11T18:45:03.092568Z",
"iopub.status.busy": "2024-01-11T18:45:03.091772Z",
"iopub.status.idle": "2024-01-11T18:45:03.146597Z",
"shell.execute_reply": "2024-01-11T18:45:03.145906Z"
},
"id": "qNLBB0zJV6vN"
},
"outputs": [
{
"data": {
"text/html": [
"\n",
" \n",
" \n",
" "
],
"text/plain": [
""
]
},
"execution_count": 27,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"Audio(wav_from_created_midi)"
]
}
],
"metadata": {
"colab": {
"collapsed_sections": [],
"name": "spice.ipynb",
"toc_visible": true
},
"kernelspec": {
"display_name": "Python 3",
"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": "3.9.18"
}
},
"nbformat": 4,
"nbformat_minor": 0
}
![](\"https://www.tensorflow.org/images/tf_logo_32px.png\"){kind=logo}
TensorFlow.orgで表示![](\"https://www.tensorflow.org/images/colab_logo_32px.png\"){kind=logo}
Google Colab で実行\n",
"![](\"https://www.tensorflow.org/images/GitHub-Mark-32px.png\")
GitHub でソースを表示![](\"https://www.tensorflow.org/images/download_logo_32px.png\"){kind=logo}
ノートブックをダウンロード![](\"https://www.tensorflow.org/images/hub_logo_32px.png\"){kind=logo}
TF Hub モデルを参照