{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 92,
   "id": "edc2e2ff",
   "metadata": {},
   "outputs": [],
   "source": [
    "import librosa\n",
    "import torch\n",
    "from transformers import Wav2Vec2Processor, HubertForCTC\n",
    "from huggingsound import SpeechRecognitionModel\n",
    "import torchaudio\n",
    "from speechbrain.pretrained import EncoderClassifier\n",
    "import time\n",
    "from transformers import Pipeline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "id": "76f25cc3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "12/06/2022 13:42:19 - INFO - huggingsound.speech_recognition.model - Loading model...\n"
     ]
    },
    {
     "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"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "12/06/2022 13:42:23 - WARNING - root - bos_token <s> not in provided tokens. It will be added to the list of tokens\n",
      "12/06/2022 13:42:23 - WARNING - root - eos_token </s> not in provided tokens. It will be added to the list of tokens\n"
     ]
    }
   ],
   "source": [
    "model_chinese = SpeechRecognitionModel(\"./wav2vec2-large-xlsr-chinese\")\n",
    "processor = Wav2Vec2Processor.from_pretrained(\"./english_fine_tune\")\n",
    "model = HubertForCTC.from_pretrained(\"./english_fine_tune\")\n",
    "language_id = EncoderClassifier.from_hparams(source=\"speechbrain/lang-id-voxlingua107-ecapa\", savedir=\"tmp\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "id": "3b142546",
   "metadata": {},
   "outputs": [],
   "source": [
    "def pipeline(path_to_audio):\n",
    "    signal = language_id.load_audio(path_to_audio)\n",
    "    prediction =  language_id.classify_batch(signal)\n",
    "    prediction[3]\n",
    "    \n",
    "    if prediction[3][0] == 'zh: Chinese':\n",
    "        print('Detected Language is Chinese')\n",
    "        transcriptions = model_chinese.transcribe([path_to_audio])\n",
    "        print(transcriptions[0]['transcription'])\n",
    "    else:\n",
    "        print('Detected language is English')\n",
    "        input_audio, sr = librosa.load(path_to_audio, sr = 16000)\n",
    "        input_values = processor(input_audio, return_tensors=\"pt\").input_values  \n",
    "        logits = model(input_values).logits\n",
    "        predicted_ids = torch.argmax(logits, dim=-1)\n",
    "        transcription = processor.decode(predicted_ids[0])\n",
    "        print(transcription)\n",
    "   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "id": "b0fae1dd",
   "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"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Detected language is English\n",
      "NISHE JUAN FANMA HE MOVED ABOUT INVISIBLE BUT EVERYONE COULD HEAR HIM\n"
     ]
    }
   ],
   "source": [
    "start = time.time()\n",
    "pipeline('combine.wav')\n",
    "end = time.time()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "id": "1e0321b5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Detected Language is Chinese\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|█████████████████████████████████████████████| 1/1 [00:00<00:00,  1.28it/s]\n",
      "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"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "你喜欢饭吗\n",
      "Detected language is English\n"
     ]
    },
    {
     "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"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Detected language is English\n"
     ]
    },
    {
     "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"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "HE MOVED ABOUT\n",
      "Detected language is English\n",
      "INVISIBLE BUT EVERYONE COULD HEAR HIM\n"
     ]
    }
   ],
   "source": [
    "from pydub import AudioSegment\n",
    "from pydub.silence import split_on_silence\n",
    "\n",
    "sound_file = AudioSegment.from_wav(\"combine.wav\")\n",
    "audio_chunks = split_on_silence(sound_file, \n",
    "    min_silence_len=100,\n",
    "    silence_thresh=-50\n",
    ")\n",
    "\n",
    "for i, chunk in enumerate(audio_chunks):\n",
    "\n",
    "    out_file = \"./chunk{0}.wav\".format(i)\n",
    "    chunk.export(out_file, format=\"wav\")\n",
    "    pipeline(out_file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a069a0fd",
   "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": "3.8.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}