{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "f3e86ccf-bfe9-48f0-a802-ff96a0bd3323",
   "metadata": {},
   "outputs": [],
   "source": [
    "import stf_alternative"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "0f33c097-c51f-4855-a632-52f1bba575f5",
   "metadata": {},
   "outputs": [],
   "source": [
    "config_path = \"front_config.json\"\n",
    "checkpoint_path = \"089.pth\"\n",
    "work_root_path = \"works\"\n",
    "device = \"cuda:0\"\n",
    "\n",
    "model = stf_alternative.create_model(\n",
    "    config_path=config_path,\n",
    "    checkpoint_path=checkpoint_path,\n",
    "    work_root_path=work_root_path,\n",
    "    device=device,\n",
    "    wavlm_path=\"microsoft/wavlm-large\",\n",
    ")\n",
    "template = stf_alternative.Template(\n",
    "    model=model,\n",
    "    config_path=config_path,\n",
    "    template_video_path=\"templates/front_one_piece_dress_nodded_cut.webm\",\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "0b4aae64-0e71-445d-9999-93f3790c61dd",
   "metadata": {},
   "outputs": [],
   "source": [
    "from pydub import AudioSegment\n",
    "silent = AudioSegment.silent(2000)\n",
    "gen_infer = template.gen_infer(audio_segment=silent,video_start_offset_frame=0)\n",
    "from PIL import Image\n",
    "for pred, chunk in gen_infer:\n",
    "    break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "fc8dec84-ebf4-4b8c-a4a3-4c2cd6fdc891",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([31, 1024])"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import torch\n",
    "model.audio_encoder(input_values=torch.randn(1, 10000).cuda()).last_hidden_state[0].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "a47a4d7d-8e46-409a-a2f6-a6b875d2c877",
   "metadata": {},
   "outputs": [],
   "source": [
    "pred = model.model(\n",
    "    img=torch.randn(1, 9, 352, 352).cuda(),\n",
    "    audio=torch.randn(1, 78, 1024).cuda(),\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "087daeec-0bc6-4ee5-97aa-ca6b4c63f3e8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([1, 3, 352, 352])"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pred.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "da6489ab-8766-4ebc-ae30-809a4a366aa2",
   "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": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}