{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "9abf3270",
   "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": 27,
   "id": "6e0830a2",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.\n"
     ]
    }
   ],
   "source": [
    "# model = AutoModelForCTC.from_pretrained(\".\").to('cuda')\n",
    "# processor = Wav2Vec2Processor.from_pretrained(\".\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "126e39e0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a2fdec3c288946a19a5b36618af4c26c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Downloading:   0%|          | 0.00/2.02k [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ae6b614b82aa4627b55b8c742ca8898e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Downloading:   0%|          | 0.00/1.18G [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "8c39ff01100c40aab1191514bb52399d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Downloading:   0%|          | 0.00/214 [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5078aede36ff4dc1a17e24a967cb46dd",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Downloading:   0%|          | 0.00/260 [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a4c7e0e6e26f4c5e8f40e79b67b61c17",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Downloading:   0%|          | 0.00/795 [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b55bcb16d8d042059a6487391e1a51de",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Downloading:   0%|          | 0.00/23.0 [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0546a666b47a4d418e62ccc8fec58bd4",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Downloading:   0%|          | 0.00/309 [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model = AutoModelForCTC.from_pretrained(\"vitouphy/xls-r-300m-km\").to('cuda')\n",
    "processor = Wav2Vec2Processor.from_pretrained(\"vitouphy/xls-r-300m-km\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "cf72163d",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using custom data configuration default-fbad308ab5a03eb2\n",
      "Reusing dataset csv (/workspace/.cache/huggingface/datasets/csv/default-fbad308ab5a03eb2/0.0.0/6b9057d9e23d9d8a2f05b985917a0da84d70c5dae3d22ddd8a3f22fb01c69d9e)\n"
     ]
    }
   ],
   "source": [
    "common_voice_test = load_dataset('csv', data_files='km_kh_male/line_index_test.csv', split = 'train')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "6b30d6ea",
   "metadata": {},
   "outputs": [],
   "source": [
    "common_voice_test  = (common_voice_test\n",
    "                      .remove_columns([\"Unnamed: 0\", \"drop\"])\n",
    "                      .rename_column('text', 'sentence'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "bf9734cc",
   "metadata": {},
   "outputs": [],
   "source": [
    "common_voice_test  = common_voice_test.cast_column(\"path\", Audio(sampling_rate=16_000)).rename_column('path', 'audio')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "5e74effa",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'audio': {'path': '/workspace/xls-r-300m-km/km_kh_male/wavs/khm_1443_3799144408.wav',\n",
       "  'array': array([-1.0600963e-06,  1.2359066e-06, -1.4001107e-06, ...,\n",
       "         -3.1423504e-05,  4.4914182e-06,  0.0000000e+00], dtype=float32),\n",
       "  'sampling_rate': 16000},\n",
       " 'sentence': 'ស៊ី ដាច់ ម៉ូតូ នៅ ពេល ដែល ប្រើ ឱ្យ ឌុប សម្ភារៈ គ្រឿង សង្ហារឹម យក ទៅ ឱ្យ ម៉ូយ នៅ ម្ដុំ វត្ដ សំរោងអណ្ដែត'}"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "common_voice_test[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "c94de1a7",
   "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": 12,
   "id": "c018376a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e9ec232c2caf4bdcb62b24696217a723",
       "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": 45,
   "id": "c0a02606",
   "metadata": {},
   "outputs": [],
   "source": [
    "i = 21"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "19b8e75f",
   "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.to(\"cuda\")).logits\n",
    "pred_ids = torch.argmax(logits, dim=-1)[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "d15fb5a5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Prediction:\n",
      "កាបូហ្វូនរ៉ាន និង កូហ្វាកទ័រ ជា មិត្ត និង គ្នា\n",
      "\n",
      "Reference:\n",
      "កាបូហ្វូនរ៉ាន និង កូហ្វាក់ទ័រ ជា មិត្ត និង គ្នា\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": "07bc1b8e",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b228faa1",
   "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
}