{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Grapheme-to-Phoneme DBP"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"\n",
"\n",
"This tutorial is available as an IPython notebook at [Malaya/example/phoneme](https://github.com/huseinzol05/Malaya/tree/master/example/phoneme).\n",
" \n",
"
"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"\n",
"\n",
"This module only trained on standard language structure, so it is not save to use it for local language structure.\n",
" \n",
"
"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"\n",
"\n",
"This module only trained on 600 samples from https://prpm.dbp.gov.my/ Glosari Dialek.\n",
" \n",
"
"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Explanation\n",
"\n",
"Phonemizer is a Grapheme-to-phoneme (G2P) conversion is the process of generating pronunciation for words based on their written form, for an example, [https://prpm.dbp.gov.my/Cari1?keyword=acaq&d=150348&#LIHATSINI](https://prpm.dbp.gov.my/Cari1?keyword=acaq&d=150348&#LIHATSINI),\n",
"\n",
"`acaq` -> `[A.tSAâÖ]`"
]
},
{
"cell_type": "code",
"execution_count": 1,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"CPU times: user 6.15 s, sys: 1.52 s, total: 7.66 s\n",
"Wall time: 11.3 s\n"
]
}
],
"source": [
"%%time\n",
"import malaya"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Use deep learning model\n",
"\n",
"Load LSTM + Bahdanau Attention phonemizer model.\n",
"\n",
"If you are using Tensorflow 2, make sure Tensorflow Addons already installed,\n",
"\n",
"```bash\n",
"pip install tensorflow-addons U\n",
"```"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"```python\n",
"def deep_model(quantized=False, **kwargs):\n",
" \"\"\"\n",
" Load LSTM + Bahdanau Attention phonetic model, \n",
" originally from https://prpm.dbp.gov.my/ Glosari Dialek.\n",
"\n",
" Original size 10.4MB, quantized size 2.77MB .\n",
"\n",
" Parameters\n",
" ----------\n",
" quantized : bool, optional (default=False)\n",
" if True, will load 8-bit quantized model.\n",
" Quantized model not necessary faster, totally depends on the machine.\n",
"\n",
" Returns\n",
" -------\n",
" result: malaya.model.tf.Seq2SeqLSTM class\n",
" \"\"\"\n",
"```"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [],
"source": [
"model = malaya.phoneme.deep_model()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Load Quantized model\n",
"\n",
"To load 8-bit quantized model, simply pass `quantized = True`, default is `False`.\n",
"\n",
"We can expect slightly accuracy drop from quantized model, and not necessary faster than normal 32-bit float model, totally depends on machine."
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"Load quantized model will cause accuracy drop.\n"
]
}
],
"source": [
"quantized_model = malaya.phoneme.deep_model(quantized = True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"#### Predict\n",
"\n",
"```python\n",
"def predict(self, strings: List[str], beam_search: bool = False):\n",
" \"\"\"\n",
" Convert to target string.\n",
"\n",
" Parameters\n",
" ----------\n",
" strings : List[str]\n",
" beam_search : bool, (optional=False)\n",
" If True, use beam search decoder, else use greedy decoder.\n",
"\n",
" Returns\n",
" -------\n",
" result: List[str]\n",
" \"\"\"\n",
"```\n",
"\n",
"If want to speed up the inference, set `beam_search = False`."
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"['sA.jA su.kA mA.kAn A.jAm', 'A.jAâÖ A.tSAâÖ kot.S\\x8d)Ö']"
]
},
"execution_count": 4,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"model.predict(['saya suka makan ayam', 'ayaq acaq kotoq'])"
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"['sA.jA su.kA mA.kAn A.jAm', 'A.jAâÖ A.tSAâÖ kot.S\\x8d)Ö']"
]
},
"execution_count": 5,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"quantized_model.predict(['saya suka makan ayam', 'ayaq acaq kotoq'])"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Limitation\n",
"\n",
"Not able to convert numbers to phoneme."
]
},
{
"cell_type": "code",
"execution_count": 10,
"metadata": {
"scrolled": true
},
"outputs": [
{
"data": {
"text/plain": [
"['A']"
]
},
"execution_count": 10,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"model.predict(['123'])"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"you have to use normalization like https://malaya.readthedocs.io/en/latest/load-num2word.html"
]
},
{
"cell_type": "code",
"execution_count": 9,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"['s«.ÒAt du.wA pu.luh ti.gA']"
]
},
"execution_count": 9,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"model.predict([malaya.num2word.to_cardinal(123)])"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"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": "3.7.7"
},
"varInspector": {
"cols": {
"lenName": 16,
"lenType": 16,
"lenVar": 40
},
"kernels_config": {
"python": {
"delete_cmd_postfix": "",
"delete_cmd_prefix": "del ",
"library": "var_list.py",
"varRefreshCmd": "print(var_dic_list())"
},
"r": {
"delete_cmd_postfix": ") ",
"delete_cmd_prefix": "rm(",
"library": "var_list.r",
"varRefreshCmd": "cat(var_dic_list()) "
}
},
"types_to_exclude": [
"module",
"function",
"builtin_function_or_method",
"instance",
"_Feature"
],
"window_display": false
}
},
"nbformat": 4,
"nbformat_minor": 4
}