{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "8a946fce-7e0d-46cb-a053-d78dc31fb121",
   "metadata": {},
   "source": [
    "## Coversion"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3c01a6d9-6569-4c94-b2a4-39696c423b1f",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "!pip install ctranslate2 transformers hf-transfer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3fd0bbb0-6496-4465-87a5-8827c1a5ed47",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "!ct2-transformers-converter \\\n",
    "    --model SadeghK/whisper-large-v3-turbo \\\n",
    "    --output_dir whisper-large-v3-turbo-ct2 \\\n",
    "    --copy_files tokenizer.json preprocessor_config.json \\\n",
    "    --quantization float16"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "b9a52b37-21d2-4758-a18f-7d59fec6fd43",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[33mWARNING: Skipping whisperx as it is not installed.\u001b[0m\u001b[33m\n",
      "\u001b[0m"
     ]
    }
   ],
   "source": [
    "!pip uninstall -y whisperx\n",
    "!rm -rf /usr/local/lib/python3.12/dist-packages/whisperx\n",
    "!rm -rf /usr/local/lib/python3.12/dist-packages/whisperx-*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "ac12e621-eec9-43b2-8186-2afc382b0ec1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "True\n"
     ]
    }
   ],
   "source": [
    "import importlib\n",
    "import sys\n",
    "print(\"whisperx\" in sys.modules)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a0f1e4c6-1b20-4ff7-9fd6-9bb16414cceb",
   "metadata": {},
   "outputs": [],
   "source": [
    "!pip install whisperx\n",
    "# !pip install git+https://github.com/m-bain/whisperx.git"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "2a98f8a1-dee5-4582-b74e-fb6e93815152",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2025-12-08 20:09:34 - whisperx.asr - INFO - No language specified, language will be detected for each audio file (increases inference time)\n",
      "2025-12-08 20:09:34 - whisperx.vads.silero - INFO - Performing voice activity detection using Silero...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using cache found in /root/.cache/torch/hub/snakers4_silero-vad_master\n"
     ]
    }
   ],
   "source": [
    "import whisperx\n",
    "device = \"cuda\"\n",
    "\n",
    "asr = whisperx.load_model(\n",
    "    \"whisper-large-v3-turbo-ct2\",\n",
    "    device=\"cuda\",\n",
    "    compute_type=\"float16\",\n",
    "    vad_method=\"silero\"\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "05764e18-df38-46d2-ae09-fbfda0a5d402",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2025-12-08 20:10:06 - whisperx.asr - WARNING - Audio is shorter than 30s, language detection may be inaccurate\n",
      "2025-12-08 20:10:06 - whisperx.asr - INFO - Detected language: fa (1.00) in first 30s of audio\n",
      "{'segments': [{'text': 'علت اصلی بارش باران جابجایی هوای مفتوب به علت اختلاف دما و رطوبت است که به جبههه\\u200cهای هوا شناسی معروف است', 'start': 0.066, 'end': 10.174}], 'language': 'fa'}\n"
     ]
    }
   ],
   "source": [
    "result = asr.transcribe(\"test.wav\")\n",
    "print(result)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "a94da22c-35a8-4e72-9897-1634e9f77998",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2b3e88bc25464c7b8efe3646fea7845a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "preprocessor_config.json:   0%|          | 0.00/158 [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d7ba2787f43a44078fcfab441e24d071",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "config.json: 0.00B [00:00, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.12/dist-packages/transformers/configuration_utils.py:335: UserWarning: Passing `gradient_checkpointing` to a config initialization is deprecated and will be removed in v5 Transformers. Using `model.gradient_checkpointing_enable()` instead, or if you are using the `Trainer` API, pass `gradient_checkpointing=True` in your `TrainingArguments`.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "817a900660334641b45cf243532c377d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "vocab.json:   0%|          | 0.00/656 [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c1f259db500240b38bdba47378988b76",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "special_tokens_map.json:   0%|          | 0.00/85.0 [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1556373079a948a8bd4543ef470a424d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "pytorch_model.bin:   0%|          | 0.00/1.26G [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "84fc59e6687447bd8e7da7ed6602106c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "model.safetensors:   0%|          | 0.00/1.26G [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "align_model, metadata = whisperx.load_align_model(\n",
    "    language_code=\"fa\",\n",
    "    device=\"cuda\"\n",
    ")\n",
    "\n",
    "aligned = whisperx.align(\n",
    "    result[\"segments\"],\n",
    "    align_model,\n",
    "    metadata,\n",
    "    \"test.wav\",\n",
    "    device=\"cuda\"\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "ff0d067b-8e66-475b-9a6d-5cf8a1c747b8",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'segments': [{'start': 0.066,\n",
       "   'end': 10.194,\n",
       "   'text': 'علت اصلی بارش باران جابجایی هوای مفتوب به علت اختلاف دما و رطوبت است که به جبههه\\u200cهای هوا شناسی معروف است',\n",
       "   'words': [{'word': 'علت',\n",
       "     'start': np.float64(0.066),\n",
       "     'end': np.float64(0.527),\n",
       "     'score': np.float64(0.926)},\n",
       "    {'word': 'اصلی',\n",
       "     'start': np.float64(0.587),\n",
       "     'end': np.float64(0.989),\n",
       "     'score': np.float64(0.945)},\n",
       "    {'word': 'بارش',\n",
       "     'start': np.float64(1.089),\n",
       "     'end': np.float64(1.47),\n",
       "     'score': np.float64(0.95)},\n",
       "    {'word': 'باران',\n",
       "     'start': np.float64(1.55),\n",
       "     'end': np.float64(1.971),\n",
       "     'score': np.float64(0.923)},\n",
       "    {'word': 'جابجایی',\n",
       "     'start': np.float64(2.453),\n",
       "     'end': np.float64(3.094),\n",
       "     'score': np.float64(0.911)},\n",
       "    {'word': 'هوای',\n",
       "     'start': np.float64(3.134),\n",
       "     'end': np.float64(3.516),\n",
       "     'score': np.float64(0.958)},\n",
       "    {'word': 'مفتوب',\n",
       "     'start': np.float64(3.596),\n",
       "     'end': np.float64(4.157),\n",
       "     'score': np.float64(0.886)},\n",
       "    {'word': 'به',\n",
       "     'start': np.float64(4.619),\n",
       "     'end': np.float64(4.679),\n",
       "     'score': np.float64(0.998)},\n",
       "    {'word': 'علت',\n",
       "     'start': np.float64(4.739),\n",
       "     'end': np.float64(5.12),\n",
       "     'score': np.float64(0.999)},\n",
       "    {'word': 'اختلاف',\n",
       "     'start': np.float64(5.18),\n",
       "     'end': np.float64(5.762),\n",
       "     'score': np.float64(0.942)},\n",
       "    {'word': 'دما',\n",
       "     'start': np.float64(5.862),\n",
       "     'end': np.float64(6.183),\n",
       "     'score': np.float64(0.822)},\n",
       "    {'word': 'و',\n",
       "     'start': np.float64(6.203),\n",
       "     'end': np.float64(6.303),\n",
       "     'score': np.float64(0.987)},\n",
       "    {'word': 'رطوبت',\n",
       "     'start': np.float64(6.343),\n",
       "     'end': np.float64(6.865),\n",
       "     'score': np.float64(0.996)},\n",
       "    {'word': 'است',\n",
       "     'start': np.float64(6.905),\n",
       "     'end': np.float64(7.126),\n",
       "     'score': np.float64(1.0)},\n",
       "    {'word': 'که',\n",
       "     'start': np.float64(7.226),\n",
       "     'end': np.float64(7.406),\n",
       "     'score': np.float64(1.0)},\n",
       "    {'word': 'به',\n",
       "     'start': np.float64(7.827),\n",
       "     'end': np.float64(7.888),\n",
       "     'score': np.float64(1.0)},\n",
       "    {'word': 'جبههه\\u200cهای',\n",
       "     'start': np.float64(7.948),\n",
       "     'end': np.float64(8.509),\n",
       "     'score': np.float64(0.804)},\n",
       "    {'word': 'هوا',\n",
       "     'start': np.float64(8.55),\n",
       "     'end': np.float64(8.79),\n",
       "     'score': np.float64(0.933)},\n",
       "    {'word': 'شناسی',\n",
       "     'start': np.float64(8.81),\n",
       "     'end': np.float64(9.332),\n",
       "     'score': np.float64(0.959)},\n",
       "    {'word': 'معروف',\n",
       "     'start': np.float64(9.392),\n",
       "     'end': np.float64(9.813),\n",
       "     'score': np.float64(0.948)},\n",
       "    {'word': 'است',\n",
       "     'start': np.float64(9.853),\n",
       "     'end': np.float64(10.194),\n",
       "     'score': np.float64(0.944)}]}],\n",
       " 'word_segments': [{'word': 'علت',\n",
       "   'start': np.float64(0.066),\n",
       "   'end': np.float64(0.527),\n",
       "   'score': np.float64(0.926)},\n",
       "  {'word': 'اصلی',\n",
       "   'start': np.float64(0.587),\n",
       "   'end': np.float64(0.989),\n",
       "   'score': np.float64(0.945)},\n",
       "  {'word': 'بارش',\n",
       "   'start': np.float64(1.089),\n",
       "   'end': np.float64(1.47),\n",
       "   'score': np.float64(0.95)},\n",
       "  {'word': 'باران',\n",
       "   'start': np.float64(1.55),\n",
       "   'end': np.float64(1.971),\n",
       "   'score': np.float64(0.923)},\n",
       "  {'word': 'جابجایی',\n",
       "   'start': np.float64(2.453),\n",
       "   'end': np.float64(3.094),\n",
       "   'score': np.float64(0.911)},\n",
       "  {'word': 'هوای',\n",
       "   'start': np.float64(3.134),\n",
       "   'end': np.float64(3.516),\n",
       "   'score': np.float64(0.958)},\n",
       "  {'word': 'مفتوب',\n",
       "   'start': np.float64(3.596),\n",
       "   'end': np.float64(4.157),\n",
       "   'score': np.float64(0.886)},\n",
       "  {'word': 'به',\n",
       "   'start': np.float64(4.619),\n",
       "   'end': np.float64(4.679),\n",
       "   'score': np.float64(0.998)},\n",
       "  {'word': 'علت',\n",
       "   'start': np.float64(4.739),\n",
       "   'end': np.float64(5.12),\n",
       "   'score': np.float64(0.999)},\n",
       "  {'word': 'اختلاف',\n",
       "   'start': np.float64(5.18),\n",
       "   'end': np.float64(5.762),\n",
       "   'score': np.float64(0.942)},\n",
       "  {'word': 'دما',\n",
       "   'start': np.float64(5.862),\n",
       "   'end': np.float64(6.183),\n",
       "   'score': np.float64(0.822)},\n",
       "  {'word': 'و',\n",
       "   'start': np.float64(6.203),\n",
       "   'end': np.float64(6.303),\n",
       "   'score': np.float64(0.987)},\n",
       "  {'word': 'رطوبت',\n",
       "   'start': np.float64(6.343),\n",
       "   'end': np.float64(6.865),\n",
       "   'score': np.float64(0.996)},\n",
       "  {'word': 'است',\n",
       "   'start': np.float64(6.905),\n",
       "   'end': np.float64(7.126),\n",
       "   'score': np.float64(1.0)},\n",
       "  {'word': 'که',\n",
       "   'start': np.float64(7.226),\n",
       "   'end': np.float64(7.406),\n",
       "   'score': np.float64(1.0)},\n",
       "  {'word': 'به',\n",
       "   'start': np.float64(7.827),\n",
       "   'end': np.float64(7.888),\n",
       "   'score': np.float64(1.0)},\n",
       "  {'word': 'جبههه\\u200cهای',\n",
       "   'start': np.float64(7.948),\n",
       "   'end': np.float64(8.509),\n",
       "   'score': np.float64(0.804)},\n",
       "  {'word': 'هوا',\n",
       "   'start': np.float64(8.55),\n",
       "   'end': np.float64(8.79),\n",
       "   'score': np.float64(0.933)},\n",
       "  {'word': 'شناسی',\n",
       "   'start': np.float64(8.81),\n",
       "   'end': np.float64(9.332),\n",
       "   'score': np.float64(0.959)},\n",
       "  {'word': 'معروف',\n",
       "   'start': np.float64(9.392),\n",
       "   'end': np.float64(9.813),\n",
       "   'score': np.float64(0.948)},\n",
       "  {'word': 'است',\n",
       "   'start': np.float64(9.853),\n",
       "   'end': np.float64(10.194),\n",
       "   'score': np.float64(0.944)}]}"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "aligned"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f8926c0e-07e6-45b7-a527-2e0aa7643bb5",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "!pip install huggingface_hub datasets --upgrade"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1e205d89-1c6c-4638-82ff-186322acac21",
   "metadata": {},
   "outputs": [],
   "source": [
    "!hf auth login --token \"<HF-TOKEN>\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6e041a99-a97f-4fcb-9a98-4d2e07991f42",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# !apt-get update && apt-get install tree"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "a15ad6d5-1db1-4ba7-8575-01099aaecaba",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[01;34m.\u001b[0m\n",
      "├── \u001b[01;35mtest.wav\u001b[0m\n",
      "├── \u001b[01;34mwhisper-large-v3-turbo-ct2\u001b[0m\n",
      "│   ├── \u001b[00mconfig.json\u001b[0m\n",
      "│   ├── \u001b[00mmodel.bin\u001b[0m\n",
      "│   ├── \u001b[00mpreprocessor_config.json\u001b[0m\n",
      "│   ├── \u001b[00mtokenizer.json\u001b[0m\n",
      "│   └── \u001b[00mvocabulary.json\u001b[0m\n",
      "└── \u001b[00mwhisper-v3-turbo-to-whisperx-pipeline.ipynb\u001b[0m\n",
      "\n",
      "2 directories, 7 files\n"
     ]
    }
   ],
   "source": [
    "!tree -L 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "d1a87396-b606-4393-9491-d11c13587be5",
   "metadata": {},
   "outputs": [],
   "source": [
    "from huggingface_hub import HfApi\n",
    "\n",
    "api = HfApi()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "a3a69f73-90e3-427c-83fd-07409fe87dac",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c187855215aa4a1aa1b3a0721d7cc43a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Processing Files (0 / 0): |          |  0.00B /  0.00B            "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c956c4df8e984234be287a31d14b1986",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "New Data Upload: |          |  0.00B /  0.00B            "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "CommitInfo(commit_url='https://huggingface.co/SadeghK/whisper-large-v3-turbo-ct2/commit/d69b61686833e10061aa5805f2d7ec33c6cf154c', commit_message='Upload folder using huggingface_hub', commit_description='', oid='d69b61686833e10061aa5805f2d7ec33c6cf154c', pr_url=None, repo_url=RepoUrl('https://huggingface.co/SadeghK/whisper-large-v3-turbo-ct2', endpoint='https://huggingface.co', repo_type='model', repo_id='SadeghK/whisper-large-v3-turbo-ct2'), pr_revision=None, pr_num=None)"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "api.upload_folder(\n",
    "    folder_path=\"whisper-large-v3-turbo-ct2\",\n",
    "    path_in_repo=\".\",\n",
    "    repo_id=\"SadeghK/whisper-large-v3-turbo-ct2\",\n",
    "    repo_type=\"model\"\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "06833856-bc89-4380-bb24-f117656a7829",
   "metadata": {},
   "source": [
    "## Inference (with fater-whisper)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "51bd55ae-4a42-4126-aeca-42ffab481d86",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "!pip install faster-whisper"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "dea2c1c8-eff2-4a21-836a-97b23eae85ec",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f4140822bc7542c09ad624b4d624fee0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "config.json: 0.00B [00:00, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3bfef789f4044b189383714c87137d03",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "model.bin:   0%|          | 0.00/1.62G [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "fe8a4e649b8640179dbb49624d06b722",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "preprocessor_config.json:   0%|          | 0.00/340 [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a3ea820e9e4c462681f28379a6cd3c39",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "tokenizer.json: 0.00B [00:00, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "210c5e1a98e749adba1ef50281027c91",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "vocabulary.json: 0.00B [00:00, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Detected language: fa\n",
      "[0.00 - 10.52] علت اصلی بارش باران جابجایی هوای مفتوب به علت اختلاف دما و رطوبت است که به جبههه‌های هوا شناسی معروف است\n"
     ]
    }
   ],
   "source": [
    "from faster_whisper import WhisperModel\n",
    "\n",
    "model = WhisperModel(\n",
    "    \"SadeghK/whisper-large-v3-turbo-ct2\",\n",
    "    device=\"cuda\",       # or \"cpu\"\n",
    "    compute_type=\"float16\"\n",
    ")\n",
    "\n",
    "segments, info = model.transcribe(\"test.wav\")\n",
    "\n",
    "print(\"Detected language:\", info.language)\n",
    "\n",
    "for seg in segments:\n",
    "    print(f\"[{seg.start:.2f} - {seg.end:.2f}] {seg.text}\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c581e58f-16a9-44cf-bb21-c11ce7f1afda",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "9096b540-da0f-43e9-9245-6acf822b17d1",
   "metadata": {},
   "source": [
    "## Inference (with whisperx)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "f43ca7e2-cae4-4d10-a8cb-7698d71cecb5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2025-12-08 20:23:55 - whisperx.asr - INFO - No language specified, language will be detected for each audio file (increases inference time)\n",
      "2025-12-08 20:23:55 - whisperx.vads.silero - INFO - Performing voice activity detection using Silero...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using cache found in /root/.cache/torch/hub/snakers4_silero-vad_master\n"
     ]
    }
   ],
   "source": [
    "import whisperx\n",
    "\n",
    "device = \"cuda\"\n",
    "\n",
    "\n",
    "asr = whisperx.load_model(\n",
    "    \"whisper-large-v3-turbo-ct2\",\n",
    "    device=\"cuda\",\n",
    "    compute_type=\"float16\",\n",
    "    vad_method=\"silero\"\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "2619bde7-b720-4457-b99c-577313b9f51a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2025-12-08 20:24:52 - whisperx.asr - WARNING - Audio is shorter than 30s, language detection may be inaccurate\n",
      "2025-12-08 20:24:52 - whisperx.asr - INFO - Detected language: fa (1.00) in first 30s of audio\n",
      "{'segments': [{'text': 'علت اصلی بارش باران جابجایی هوای مفتوب به علت اختلاف دما و رطوبت است که به جبههه\\u200cهای هوا شناسی معروف است', 'start': 0.066, 'end': 10.174}], 'language': 'fa'}\n"
     ]
    }
   ],
   "source": [
    "audio_file = \"test.wav\"\n",
    "asr_result = asr.transcribe(audio_file)\n",
    "\n",
    "print(asr_result)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "312b66a1-e900-44e5-965f-1dbb139c5702",
   "metadata": {},
   "source": [
    "## Inference (with whisperx + word-lvl alignment)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "f030a2b5-4314-4a49-8991-7055d249a71e",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2025-12-08 20:34:40 - whisperx.asr - INFO - No language specified, language will be detected for each audio file (increases inference time)\n",
      "2025-12-08 20:34:40 - whisperx.vads.silero - INFO - Performing voice activity detection using Silero...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using cache found in /root/.cache/torch/hub/snakers4_silero-vad_master\n"
     ]
    }
   ],
   "source": [
    "## \n",
    "import whisperx\n",
    "\n",
    "device = \"cuda\"\n",
    "\n",
    "# 1) Load ASR model\n",
    "asr = whisperx.load_model(\n",
    "    \"SadeghK/whisper-large-v3-turbo-ct2\",\n",
    "    device=device,\n",
    "    compute_type=\"float16\",\n",
    "    vad_method=\"silero\"\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "d7cdfc4a-c737-4204-b0f7-6661df119de8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2025-12-08 20:34:46 - whisperx.asr - WARNING - Audio is shorter than 30s, language detection may be inaccurate\n",
      "2025-12-08 20:34:46 - whisperx.asr - INFO - Detected language: fa (1.00) in first 30s of audio\n",
      "ASR: {'segments': [{'text': 'علت اصلی بارش باران جابجایی هوای مفتوب به علت اختلاف دما و رطوبت است که به جبههه\\u200cهای هوا شناسی معروف است', 'start': 0.066, 'end': 10.174}], 'language': 'fa'}\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.12/dist-packages/transformers/configuration_utils.py:335: UserWarning: Passing `gradient_checkpointing` to a config initialization is deprecated and will be removed in v5 Transformers. Using `model.gradient_checkpointing_enable()` instead, or if you are using the `Trainer` API, pass `gradient_checkpointing=True` in your `TrainingArguments`.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[{'start': 0.066, 'end': 10.194, 'text': 'علت اصلی بارش باران جابجایی هوای مفتوب به علت اختلاف دما و رطوبت است که به جبههه\\u200cهای هوا شناسی معروف است', 'words': [{'word': 'علت', 'start': np.float64(0.066), 'end': np.float64(0.527), 'score': np.float64(0.926)}, {'word': 'اصلی', 'start': np.float64(0.587), 'end': np.float64(0.989), 'score': np.float64(0.945)}, {'word': 'بارش', 'start': np.float64(1.089), 'end': np.float64(1.47), 'score': np.float64(0.95)}, {'word': 'باران', 'start': np.float64(1.55), 'end': np.float64(1.971), 'score': np.float64(0.923)}, {'word': 'جابجایی', 'start': np.float64(2.453), 'end': np.float64(3.094), 'score': np.float64(0.911)}, {'word': 'هوای', 'start': np.float64(3.134), 'end': np.float64(3.516), 'score': np.float64(0.958)}, {'word': 'مفتوب', 'start': np.float64(3.596), 'end': np.float64(4.157), 'score': np.float64(0.886)}, {'word': 'به', 'start': np.float64(4.619), 'end': np.float64(4.679), 'score': np.float64(0.998)}, {'word': 'علت', 'start': np.float64(4.739), 'end': np.float64(5.12), 'score': np.float64(0.999)}, {'word': 'اختلاف', 'start': np.float64(5.18), 'end': np.float64(5.762), 'score': np.float64(0.942)}, {'word': 'دما', 'start': np.float64(5.862), 'end': np.float64(6.183), 'score': np.float64(0.822)}, {'word': 'و', 'start': np.float64(6.203), 'end': np.float64(6.303), 'score': np.float64(0.987)}, {'word': 'رطوبت', 'start': np.float64(6.343), 'end': np.float64(6.865), 'score': np.float64(0.996)}, {'word': 'است', 'start': np.float64(6.905), 'end': np.float64(7.126), 'score': np.float64(1.0)}, {'word': 'که', 'start': np.float64(7.226), 'end': np.float64(7.406), 'score': np.float64(1.0)}, {'word': 'به', 'start': np.float64(7.827), 'end': np.float64(7.888), 'score': np.float64(1.0)}, {'word': 'جبههه\\u200cهای', 'start': np.float64(7.948), 'end': np.float64(8.509), 'score': np.float64(0.804)}, {'word': 'هوا', 'start': np.float64(8.55), 'end': np.float64(8.79), 'score': np.float64(0.933)}, {'word': 'شناسی', 'start': np.float64(8.81), 'end': np.float64(9.332), 'score': np.float64(0.959)}, {'word': 'معروف', 'start': np.float64(9.392), 'end': np.float64(9.813), 'score': np.float64(0.948)}, {'word': 'است', 'start': np.float64(9.853), 'end': np.float64(10.194), 'score': np.float64(0.944)}]}]\n"
     ]
    }
   ],
   "source": [
    "audio = \"test.wav\"\n",
    "asr_result = asr.transcribe(audio)\n",
    "print(\"ASR:\", asr_result)\n",
    "\n",
    "# 2) Load alignment model (Persian = 'fa')\n",
    "align_model, metadata = whisperx.load_align_model(\n",
    "    language_code=\"fa\",\n",
    "    device=device\n",
    ")\n",
    "\n",
    "# 3) Run alignment\n",
    "aligned = whisperx.align(\n",
    "    asr_result[\"segments\"],\n",
    "    align_model,\n",
    "    metadata,\n",
    "    audio,\n",
    "    device=device\n",
    ")\n",
    "\n",
    "print(aligned[\"segments\"])     # includes word-level timestamps\n"
   ]
  }
 ],
 "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": "3.12.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}