{
 "cells": [
  {
   "metadata": {},
   "cell_type": "markdown",
   "source": "Audio",
   "id": "8b8c1a352260e82a"
  },
  {
   "cell_type": "code",
   "id": "initial_id",
   "metadata": {
    "collapsed": true,
    "ExecuteTime": {
     "end_time": "2025-06-10T09:38:10.760409Z",
     "start_time": "2025-06-10T09:38:10.617508Z"
    }
   },
   "source": [
    "from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor\n",
    "import torch\n",
    "import torchaudio.transforms as T\n",
    "import pydub\n",
    "import numpy as np"
   ],
   "outputs": [],
   "execution_count": 5
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-06-10T09:43:53.684713Z",
     "start_time": "2025-06-10T09:43:53.681866Z"
    }
   },
   "cell_type": "code",
   "source": [
    "# CONSTANTS\n",
    "audio_model_dir = './models_for_proj/wav2vec2-base-960h'\n",
    "\n",
    "# audio_dir = 'files/1f975693-876d-457b-a649-393859e79bf3.mp3'\n",
    "audio_dir = 'files/99c9cc74-fdc8-46c6-8f8d-3ce2d3bfeea3.mp3'"
   ],
   "id": "3ee50d096b2c9d44",
   "outputs": [],
   "execution_count": 19
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-06-10T09:43:54.053411Z",
     "start_time": "2025-06-10T09:43:54.006676Z"
    }
   },
   "cell_type": "code",
   "source": [
    "\n",
    "model = Wav2Vec2ForCTC.from_pretrained(audio_model_dir)\n",
    "processor = Wav2Vec2Processor.from_pretrained(audio_model_dir)"
   ],
   "id": "b51a485af7b9cf14",
   "outputs": [],
   "execution_count": 20
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-06-10T09:43:54.603559Z",
     "start_time": "2025-06-10T09:43:54.414677Z"
    }
   },
   "cell_type": "code",
   "source": [
    "def read_mp3(f, normalized=False):\n",
    "    \"\"\"Read MP3 file to numpy array.\"\"\"\n",
    "    a = pydub.AudioSegment.from_mp3(f)\n",
    "    y = np.array(a.get_array_of_samples())\n",
    "    if a.channels == 2:\n",
    "        y = y.reshape((-1, 2))\n",
    "    if normalized:\n",
    "        return a.frame_rate, np.float32(y) / 2**15\n",
    "    else:\n",
    "        return a.frame_rate, y\n",
    "\n",
    "# Usage\n",
    "audio_input_sr, audio_input_np = read_mp3(audio_dir)"
   ],
   "id": "ac7e2b43ace4d232",
   "outputs": [],
   "execution_count": 21
  },
  {
   "metadata": {
    "ExecuteTime": {
     "end_time": "2025-06-10T09:43:56.920665Z",
     "start_time": "2025-06-10T09:43:56.244101Z"
    }
   },
   "cell_type": "code",
   "source": [
    "# --------------------------------------------------------------------------- #\n",
    "# audio_input_sr, audio_input_np = audio_input\n",
    "audio_input_t = torch.tensor(audio_input_np, dtype=torch.float32)\n",
    "target_sr = 16000\n",
    "resampler = T.Resample(audio_input_sr, target_sr, dtype=audio_input_t.dtype)\n",
    "resampled_audio_input_t: torch.Tensor = resampler(audio_input_t)\n",
    "resampled_audio_input_np = resampled_audio_input_t.numpy()\n",
    "# --------------------------------------------------------------------------- #\n",
    "# result = asr_pipe_default(resampled_audio_input_np)\n",
    "inputs = processor(resampled_audio_input_np, sampling_rate=16000, return_tensors=\"pt\", padding=True)\n",
    "# Inference\n",
    "with torch.no_grad():\n",
    "    logits = model(**inputs).logits\n",
    "# Decode\n",
    "predicted_ids = torch.argmax(logits, dim=-1)\n",
    "transcription = processor.decode(predicted_ids[0])\n",
    "# print(\"Transcription:\", transcription)\n",
    "transcription"
   ],
   "id": "2a4738e9d038985",
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'IN A SAUCEPAN COMBINE RIPE STRAWBERRIES GRANULATED SUGAR FRESHLY SQUEEZED LEMON JUICE AND CORNSTARCH COOK THE MIXTURE OF A MEDIUM HEAT STIRRING CONSTANTLY UNTIL IT THICKENS TO A SMOOTH CONSISTENCY REMOVE FROM HEAT AND STIR IN A DASH OF PURE VANILLA EXTRACT ALLOW THE STRAWBERRY PIE FEELING TO COOL BEFORE USING IT AS A DELICIOUS AND FRUITY FILLING FOR YOUR PIE CRUST'"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "execution_count": 22
  },
  {
   "metadata": {},
   "cell_type": "code",
   "outputs": [],
   "execution_count": null,
   "source": "",
   "id": "f159c2955f140600"
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}