{
"cells": [
{
"cell_type": "code",
"execution_count": 1,
"id": "33ff07b7",
"metadata": {},
"outputs": [],
"source": [
"import json\n",
"import soundfile as sf\n",
"from tqdm import tqdm"
]
},
{
"cell_type": "code",
"execution_count": 2,
"id": "8b36e5d6",
"metadata": {},
"outputs": [],
"source": [
"!mkdir audio"
]
},
{
"cell_type": "code",
"execution_count": 3,
"id": "26f935d3",
"metadata": {},
"outputs": [],
"source": [
"with open('/home/husein/speech-bahasa/malay-asr-train-shuffled.json') as fopen:\n",
" data = json.load(fopen)"
]
},
{
"cell_type": "code",
"execution_count": 6,
"id": "c9d2279d",
"metadata": {},
"outputs": [],
"source": [
"# !wget https://gist.githubusercontent.com/huseinzol05/98974ae8c6c7a65d4bc0af9f5003786a/raw/5aa5257608b61e8fcc828e99fbd070d5ca7358e3/mp.py"
]
},
{
"cell_type": "code",
"execution_count": 7,
"id": "51e46ebc",
"metadata": {},
"outputs": [],
"source": [
"import mp\n",
"\n",
"def loop(index):\n",
" index, _ = index\n",
" for i in tqdm(index):\n",
" y, _ = sf.read(data['X'][i])\n",
" sf.write(f'audio/{i}.mp3', y, 16000)"
]
},
{
"cell_type": "code",
"execution_count": 8,
"id": "61fb2fd6",
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|███████████████████████████████████████████| 10/10 [00:02<00:00, 4.63it/s]\n"
]
}
],
"source": [
"loop((range(10), 0))"
]
},
{
"cell_type": "code",
"execution_count": 9,
"id": "45421d6d",
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
" 76%|████████████████████████▎ | 61997/81779 [20:02:48<7:46:14, 1.41s/it]IOPub message rate exceeded.\n",
"The notebook server will temporarily stop sending output\n",
"to the client in order to avoid crashing it.\n",
"To change this limit, set the config variable\n",
"`--NotebookApp.iopub_msg_rate_limit`.\n",
"\n",
"Current values:\n",
"NotebookApp.iopub_msg_rate_limit=1000.0 (msgs/sec)\n",
"NotebookApp.rate_limit_window=3.0 (secs)\n",
"\n",
"100%|██████████████████████████████████| 81779/81779 [26:37:45<00:00, 1.17s/it]\n",
"100%|██████████████████████████████████| 81779/81779 [26:37:49<00:00, 1.17s/it]\n",
"100%|██████████████████████████████████| 81779/81779 [26:37:51<00:00, 1.17s/it]\n",
"100%|███████████████████████████████████████████| 19/19 [00:23<00:00, 1.22s/it]\n",
"100%|██████████████████████████████████| 81779/81779 [26:38:09<00:00, 1.17s/it]\n",
"100%|██████████████████████████████████| 81779/81779 [26:38:16<00:00, 1.17s/it]\n",
"100%|██████████████████████████████████| 81779/81779 [26:38:17<00:00, 1.17s/it]\n",
"100%|██████████████████████████████████| 81779/81779 [26:38:17<00:00, 1.17s/it]\n",
"100%|██████████████████████████████████| 81779/81779 [26:38:21<00:00, 1.17s/it]\n",
"100%|██████████████████████████████████| 81779/81779 [26:38:29<00:00, 1.17s/it]\n",
"100%|██████████████████████████████████| 81779/81779 [26:38:57<00:00, 1.17s/it]\n",
"100%|██████████████████████████████████| 81779/81779 [26:39:09<00:00, 1.17s/it]\n",
"100%|██████████████████████████████████| 81779/81779 [26:39:09<00:00, 1.17s/it]\n",
"100%|██████████████████████████████████| 81779/81779 [26:39:21<00:00, 1.17s/it]\n",
"100%|██████████████████████████████████| 81779/81779 [26:39:30<00:00, 1.17s/it]\n",
"100%|██████████████████████████████████| 81779/81779 [26:39:33<00:00, 1.17s/it]\n",
"100%|██████████████████████████████████| 81779/81779 [26:39:48<00:00, 1.17s/it]\n",
"100%|██████████████████████████████████| 81779/81779 [26:39:52<00:00, 1.17s/it]\n",
"100%|██████████████████████████████████| 81779/81779 [26:39:56<00:00, 1.17s/it]\n",
"100%|██████████████████████████████████| 81779/81779 [26:40:06<00:00, 1.17s/it]\n",
"100%|██████████████████████████████████| 81779/81779 [26:40:12<00:00, 1.17s/it]\n"
]
}
],
"source": [
"mp.multiprocessing(range(len(data['X'])), loop, cores = 20, returned = False)"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "deb07990",
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"100%|█████████████████████████████| 1635599/1635599 [00:03<00:00, 484512.06it/s]\n"
]
}
],
"source": [
"import os\n",
"\n",
"with open('label.jsonl', 'w') as fopen:\n",
" for i in tqdm(range(len(data['X']))):\n",
" f = f'audio/{i}.mp3'\n",
" d = {\n",
" 'filename': f,\n",
" 'Y': data['Y'][i]\n",
" }\n",
" fopen.write(f'{json.dumps(d)}\\n')"
]
},
{
"cell_type": "code",
"execution_count": 24,
"id": "85353fef",
"metadata": {},
"outputs": [],
"source": [
"from datasets import load_dataset, Audio"
]
},
{
"cell_type": "code",
"execution_count": 25,
"id": "5676f8b5",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Downloading and preparing dataset json/default to /home/husein/.cache/huggingface/datasets/json/default-e2b95159dbdc9fa6/0.0.0/e347ab1c932092252e717ff3f949105a4dd28b27e842dd53157d2f72e276c2e4...\n"
]
},
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "589c88f771424543b3d60f24f89fbe83",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
"Downloading data files: 0%| | 0/1 [00:00\n",
" \n",
" Your browser does not support the audio element.\n",
" \n",
" "
],
"text/plain": [
""
]
},
"execution_count": 38,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"import IPython.display as ipd\n",
"ipd.Audio(dataset[10000]['filename']['array'], rate = 16000)"
]
},
{
"cell_type": "code",
"execution_count": 40,
"id": "390a8aff",
"metadata": {},
"outputs": [
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "26d553fe2c7a485fba88312495a124e8",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
"Map: 0%| | 0/15286 [00:00