{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "1b285579",
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformers import AutoModelForCTC, Wav2Vec2Processor\n",
    "from datasets import load_dataset, load_metric, Audio\n",
    "import numpy as np\n",
    "import torch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "1f3354ce",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = AutoModelForCTC.from_pretrained(\".\")\n",
    "processor = Wav2Vec2Processor.from_pretrained(\".\")\n",
    "\n",
    "# model = AutoModelForCTC.from_pretrained(\"vitouphy/wav2vec2-xls-r-1b-km\")\n",
    "# processor = Wav2Vec2Processor.from_pretrained(\"vitouphy/wav2vec2-xls-r-1b-km\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "bdcb886d",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using custom data configuration default-36119ec2a15afb82\n",
      "Reusing dataset csv (/workspace/.cache/huggingface/datasets/csv/default-36119ec2a15afb82/0.0.0/6b9057d9e23d9d8a2f05b985917a0da84d70c5dae3d22ddd8a3f22fb01c69d9e)\n"
     ]
    }
   ],
   "source": [
    "common_voice_test  = (load_dataset('csv', data_files='/workspace/xls-r-300m-km/km_kh_male/line_index_test.csv', split = 'train')\n",
    "                      .remove_columns([\"Unnamed: 0\", \"drop\"])\n",
    "                      .rename_column('text', 'sentence')\n",
    "                      .cast_column(\"path\", Audio(sampling_rate=16_000)).rename_column('path', 'audio'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "e6631b64",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'audio': {'path': '/workspace/xls-r-300m-km/km_kh_male/wavs/khm_3154_2555595821.wav',\n",
       "  'array': array([ 0.00014737,  0.00016698,  0.00013704, ..., -0.00011244,\n",
       "         -0.0001059 , -0.00011476], dtype=float32),\n",
       "  'sampling_rate': 16000},\n",
       " 'sentence': 'ការ ធ្វើ អាជីវកម្ម រ៉ែ ដំបូង នៅ កម្ពុជា'}"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "common_voice_test[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "6357f917",
   "metadata": {},
   "outputs": [],
   "source": [
    "def prepare_dataset(batch):\n",
    "    audio = batch[\"audio\"]\n",
    "    \n",
    "    # batched output is \"un-batched\"\n",
    "    batch[\"input_values\"] = processor(np.array(audio[\"array\"]), sampling_rate=audio[\"sampling_rate\"]).input_values[0]\n",
    "    batch[\"input_length\"] = len(batch[\"input_values\"])\n",
    "    \n",
    "    with processor.as_target_processor():\n",
    "        batch[\"labels\"] = processor(batch[\"sentence\"]).input_ids\n",
    "    return batch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "86b15f84",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d422eded8090415fa2af118843279369",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "0ex [00:00, ?ex/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "common_voice_test = common_voice_test.map(prepare_dataset, remove_columns=common_voice_test.column_names)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "d29cda20",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "72"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "processor.tokenizer.pad_token_id"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "2b38bde0",
   "metadata": {},
   "outputs": [],
   "source": [
    "i = 25"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "7e22806f",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "It is strongly recommended to pass the ``sampling_rate`` argument to this function. Failing to do so can result in silent errors that might be hard to debug.\n"
     ]
    }
   ],
   "source": [
    "input_dict = processor(common_voice_test[i][\"input_values\"], return_tensors=\"pt\", padding=True)\n",
    "logits = model(input_dict.input_values).logits\n",
    "pred_ids = torch.argmax(logits, dim=-1)[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "c804d1e6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Prediction:\n",
      "កញ្ញាទេ បូព្រឹក សម្ដែង នៅ តន្ត្រី ស្រាប៊ៀរ កម្ពុជា\n",
      "\n",
      "Reference:\n",
      "កញ្ញា ទេព បូព្រឹក្ស សម្ដែង នៅ តន្ត្រី ស្រាបៀរ កម្ពុជា\n"
     ]
    }
   ],
   "source": [
    "print(\"Prediction:\")\n",
    "pred_ids = pred_ids[pred_ids != processor.tokenizer.pad_token_id]\n",
    "print(processor.decode(pred_ids))\n",
    "\n",
    "print(\"\\nReference:\")\n",
    "print(processor.decode(common_voice_test['labels'][i]))\n",
    "# print(common_voice_test_transcription[0][\"sentence\"].lower())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7c576143",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5745b14f",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}