{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "%matplotlib inline"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "\n# Filter Out Background Noise\n\n\nFilter out background noise from noisy speech signals. \n\nTo see how soundpy implements this, see `soundpy.builtin.filtersignal`.\n\nAs a general note for filtering, the Wiener Filter is the default filter for soundpy. It seems to filter signals more consequently than the Band Spectral Subtraction Filter.\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "# Let's import soundpy, and ipd for playing audio data\nimport soundpy as sp\nimport IPython.display as ipd"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Define the noisy and clean speech audio files.\n^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nNote: these files are available in the soundpy repo.\nDesignate path relevant for accessing audiodata\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "sp_dir = '../../../'"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Noise sample:\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "noise = '{}audiodata/background_samples/traffic.wav'.format(sp_dir)\nnoise = sp.string2pathlib(noise)\nspeech = '{}audiodata/python.wav'.format(sp_dir)\nspeech = sp.utils.string2pathlib(speech)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "For filtering, we will set the sample rate to be quite high:\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "sr = 48000"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Create noisy speech signal as SNR 10\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "noisy, snr_measured = sp.dsp.add_backgroundsound(\n    speech, \n    noise, \n    sr = sr, \n    snr = 10, \n    total_len_sec = 2, \n    pad_mainsound_sec = 0.5)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Hear and see the noisy speech \n~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "ipd.Audio(noisy,rate=sr)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "sp.plotsound(noisy, sr=sr, feature_type='signal', \n               title = 'Noisy Speech', subprocess=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Hear and see the clean speech \n~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "s, sr = sp.loadsound(speech, sr=sr)\nipd.Audio(s,rate=sr)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "sp.plotsound(s, sr=sr, feature_type='signal', \n               title = 'Clean Speech', subprocess=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Filter the noisy speech\n^^^^^^^^^^^^^^^^^^^^^^^\n\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Wiener Filter \n~~~~~~~~~~~~~\n\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Let's filter with a Wiener filter:\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "noisy_wf, sr = sp.filtersignal(noisy,\n                                 sr = sr,\n                                 filter_type = 'wiener') # default"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "ipd.Audio(noisy_wf,rate=sr)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "sp.plotsound(noisy_wf, sr = sr, feature_type = 'signal', \n               title = 'Noisy Speech: Wiener Filter', \n               subprocess=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Wiener Filter with Postfilter\n~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Let's filter with a Wiener filter and postfilter\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "noisy_wfpf, sr = sp.filtersignal(noisy,\n                                 sr = sr,\n                                 filter_type = 'wiener',\n                                 apply_postfilter = True)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "ipd.Audio(noisy_wfpf,rate=sr)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "sp.plotsound(noisy_wfpf, sr=sr, feature_type = 'signal', \n               title = 'Noisy Speech: Wiener Filter with Postfilter', \n               subprocess=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Band Spectral Subtraction\n~~~~~~~~~~~~~~~~~~~~~~~~~\n\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Let's filter using band spectral subtraction\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "noisy_bs, sr = sp.filtersignal(noisy,\n                                 sr = sr,\n                                 filter_type = 'bandspec')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "ipd.Audio(noisy_bs,rate=sr)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "sp.plotsound(noisy_bs, sr = sr, feature_type = 'signal', \n               title = 'Noisy Speech: Band Spectral Subtraction', \n               subprocess=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Band Spectral Subtraction with Postfilter\n~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Finally, let's filter using band spectral subtraction with a postfilter\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "noisy_bspf, sr = sp.filtersignal(noisy,\n                                 sr = sr,\n                                 filter_type = 'bandspec', \n                                 apply_postfilter = True)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "ipd.Audio(noisy_bspf,rate=sr)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "sp.plotsound(noisy_bspf, sr = sr, feature_type = 'signal', \n               title = 'Noisy Speech: Band Spectral Subtraction with Postfilter', \n               subprocess=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Filter: increase the scale\n^^^^^^^^^^^^^^^^^^^^^^^^^^\n\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Let's filter with a Wiener filter:\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "filter_scale = 5\nnoisy_wf, sr = sp.filtersignal(noisy,\n                                 sr=sr,\n                                 filter_type = 'wiener',\n                                 filter_scale = filter_scale)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Wiener Filter\n~~~~~~~~~~~~~\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "ipd.Audio(noisy_wf,rate=sr)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "sp.plotsound(noisy_wf, sr = sr, feature_type = 'signal', \n               title = 'Noisy Speech: Wiener Filter Scale {}'.format(filter_scale), \n               subprocess=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Wiener Filter with Postfilter\n~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "Let's filter with a Wiener filter and postfilter\n\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "noisy_wfpf, sr = sp.filtersignal(noisy,\n                                 sr = sr,\n                                 filter_type = 'wiener',\n                                 apply_postfilter = True,\n                                 filter_scale = filter_scale)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "ipd.Audio(noisy_wfpf,rate = sr)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "collapsed": false
      },
      "outputs": [],
      "source": [
        "sp.plotsound(noisy_wfpf, sr = sr, feature_type = 'signal', \n               title = 'Noisy Speech: Wiener Filter with Postfilter Scale {}'.format(filter_scale),\n               subprocess=True)"
      ]
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3",
      "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": "3.8.2"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}