{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import wave\n",
    "\n",
    "from librosa import resample\n",
    "from IPython.display import Audio\n",
    "from transformers import pipeline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def open_wave(wav_filename):\n",
    "  with wave.open(wav_filename, mode=\"rb\") as wav_in:\n",
    "    if wav_in.getsampwidth() != 2:\n",
    "      raise Exception(\"Input not 16-bit\")\n",
    "\n",
    "    nchannels = wav_in.getnchannels()\n",
    "    nframes = wav_in.getnframes()\n",
    "    nsamples = nchannels * nframes\n",
    "    xb = wav_in.readframes(nframes)\n",
    "    b_np = np.frombuffer(xb, dtype=np.int16) / nchannels\n",
    "    samples = [int(sum(b_np[b0 : b0 + nchannels])) for b0 in range(0, nsamples, nchannels)]\n",
    "\n",
    "    return (samples, wav_in.getframerate())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "pipe = pipeline(\"automatic-speech-recognition\", model=\"openai/whisper-base\", chunk_length_s=30)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def transcribe(samples, orig_sr=44100, target_sr=16000):\n",
    "  min_s, max_s = min(samples), max(samples)\n",
    "  samples_f = 2.0 * (np.array(samples) - min_s) / (max_s - min_s) - 1.0\n",
    "  resamples = resample(samples_f, orig_sr=orig_sr, target_sr=target_sr)\n",
    "  prediction = pipe(resamples.copy(), batch_size=8)\n",
    "  return prediction[\"text\"].strip().lower()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "samples, sr = open_wave(\"./audio/plain_01.wav\")\n",
    "display(Audio(samples, rate=sr))\n",
    "transcribe(samples, sr)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "gradio",
   "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": "3.9.18"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}