"2. 1 test set is tested using the classifier trained on the remaining 9.\n",
"3. We then do test/train on all of the other sets and average the percentages. \n",
"\n",
"To achieve the first step (divide our training set into k disjoint subsets), use the function [Kfold](http://scikit-learn.org/stable/modules/generated/sklearn.cross_validation.KFold.html) in the scikit.learn cross_validation package.\n",
"To achieve the first step (divide our training set into k disjoint subsets), use the function [Kfold](http://scikit-learn.org/stable/modules/generated/sklearn.cross_validation.KFold.html) in the scikit.learn cross_validation package:\n",
"\n",
" K-Folds cross validation iterator.\n",
" Provides train/test indices to split data in train test sets. Split dataset into k consecutive folds (without shuffling).\n",
"\n",
" You can visit the scikit.learn documentation to look at all the other options. This code is also posted as a template in \n",
"My first audio classifier: introducing K-NN! \n",
"\n",
"We can now appreciate why we need additional intelligence in our systems - heuristics can't go very far in the world of complex audio signals. We'll be using scikit.learn's implementation of the k-NN for our work here. It proves be a straight-forward and easy to use implementation. The steps and skills of working with one classifier will scale nicely to working with other, more complex classifiers. \n",
"\n",
"We're also going to be using the new features in our arsenal: cherishing those \"spectral moments\" (centroid, bandwidth, skewness, kurtosis) and also examining other spectral statistics. \n",
" \n",
"### Training Data\n",
"\n",
"First off, we want to analyze and feature extract a small collection of audio samples - storing their feature data as our \"training data\". The commands below read all of the drum example .wav files from the MIR web site into an array, snareFileList. \n",
"\n",
"First we define a function to retrieve a list of URLs from a text file."
]
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"import urllib2\n",
"\n",
"def process_corpus(corpus_URL):\n",
" \"\"\"Read a list of files to process from the text file at corpusURL. Return a list of URLs\"\"\" \n",
" # Open and read each line\n",
" url_list_text_data = urllib2.urlopen(corpus_URL) # it's a file like object and works just like a file\n",
" for file_URL in url_list_text_data: # files are iterable\n",
" yield file_URL.rstrip()"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 1
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Use these commands to read in a list of filenames (samples) in a directory, replacing the URL with a URL to a list of URLs (one per line) indicating where the audio / drum samples are stored."
"kick_file_list = [audio_file_URL for audio_file_URL in process_corpus(kicks_URL)]"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 3
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"To access the filenames contained in the array, use the square brackets [ ] to get to the element that you want to access. For example, to access the text URL file name of the first file in the list, you would type:"
"When we feature extract a sample collection, we need to sequentially access audio files, segment them (or not), and feature extract them. Loading a lot of audio files into memory is not always a feasible or desirable operation, so you will create a loop which loads an audio file, feature extracts it, and closes the audio file. Note that the only information that we retain in memory are the features that are extracted.\n",
"\n",
"Create a loop which reads in an audio file, extracts the zero crossing rate, and some spectral statistics. You can use the \"in\" operator to retrieve each audio file URL from process_corpus(), as used above. The feature information for each audio file (the \"feature vector\") should be stored as a feature array, with columns being the features and rows for each file. For example:\n",
"\n",
" featuresSnare =\n",
"\n",
" 0.5730 1.9183 2.9713 0.0004 0.0002\n",
" 0.4750 1.4834 2.4463 0.0004 0.0012\n",
" 0.5900 2.2857 3.1788 0.0003 0.0041\n",
" 0.5090 1.6622 2.6369 0.0004 0.0051\n",
" 0.4860 1.4758 2.2085 0.0004 0.0021\n",
" 0.6060 2.2119 3.2798 0.0004 0.0651\n",
" 0.4990 2.0607 2.7654 0.0004 0.0721\n",
" 0.6360 2.3153 3.0256 0.0003 0.0221\n",
" 0.5490 2.0137 3.0342 0.0004 0.0016\n",
" 0.5900 2.2857 3.1788 0.0003 0.0012\n",
" \n",
" Within your loop, here's a reminder how to read in your wav files, using an array of audio file URLs:"
"4. First, extract all of the feature data for the kick drums and store it in a feature array. (For my example, above, I'd put it in \"featuresKick\")\n",
"\n",
"5. Next, extract all of the feature data for the snares, storing them in a different array. \n",
"Again, the kick and snare features should be separated in two different arrays!\n",
" \n",
"OK, no more help. The rest is up to you!"
"We can now appreciate why we need additional intelligence in our systems - heuristics can't go very far in the world of complex audio signals. We'll be using scikit.learn's implementation of the k-NN for our work here. It proves be a straight-forward and easy to use implementation. The steps and skills of working with one classifier will scale nicely to working with other, more complex classifiers."
]
},
{
...
...
@@ -189,7 +30,7 @@
"\n",
"2. Since the features are different scales, we will want to normalize each feature vector to a common range - storing the scaling coefficients for later use. Many techniques exist for scaling your features. We'll use linear scaling, which forces the features into the range -1 to 1.\n",
"\n",
" For this, we'll use a scikit.learn class called [MinMaxScaler](http://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.MinMaxScaler.html). MinMaxScaler fits and transforms, returning an array of scaled values, and retains coefficients which were used to scale each column into -1 to 1. Use these functions in your code."
" For this, we'll use a scikit.learn class called [MinMaxScaler](http://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.MinMaxScaler.html). MinMaxScaler fits and transforms, returning an array of scaled values, and retains coefficients which were used to scale each column into -1 to 1. Use these functions in your code."
"\u001b[0;32m<ipython-input-4-a41b13b7e64c>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mmodel_snare\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mtrainingFeatures\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlabels\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
"\u001b[0;31mNameError\u001b[0m: name 'trainingFeatures' is not defined"
]
}
],
"prompt_number": 4
},
{
"cell_type": "markdown",
...
...
@@ -298,7 +152,7 @@
" \n",
"### Rescaling\n",
"\n",
"In evaluating a new audio file, we need to extract it's features, re-scale them to the same range as the trained feature values, and then send them through the knn."
"In evaluating a new audio file, we need to extract it's features, re-scale them to the same range as the trained feature values, and then send them through the k-NN."
"\u001b[0;32m<ipython-input-12-3e535d776ccc>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m 1\u001b[0m \u001b[0;31m# This uses the previous calculated linear scaling parameters to adjust the incoming features to the same range.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m \u001b[0mfeaturesScaled\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mscaler\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtransform\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfeatures\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
"\u001b[0;32m<ipython-input-5-3e535d776ccc>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m 1\u001b[0m \u001b[0;31m# This uses the previous calculated linear scaling parameters to adjust the incoming features to the same range.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m \u001b[0mfeaturesScaled\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mscaler\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtransform\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfeatures\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
"\u001b[0;31mNameError\u001b[0m: name 'features' is not defined"
"For classification, we're going to be using the new features in our arsenal: cherishing those \"spectral moments\" (centroid, bandwidth, skewness, kurtosis) and also examining other spectral statistics."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Training Data\n",
"\n",
"First off, we want to analyze and feature extract a small collection of audio samples - storing their feature data as our \"training data\". The commands below read all of the drum example .wav files from the MIR web site into an array, snareFileList. \n",
"\n",
"First we define a function to retrieve a list of URLs from a text file."
]
},
{
"cell_type": "code",
"collapsed": false,
"input": [
"import urllib2\n",
"\n",
"def process_corpus(corpus_URL):\n",
" \"\"\"Read a list of files to process from the text file at corpusURL. Return a list of URLs\"\"\" \n",
" # Open and read each line\n",
" url_list_text_data = urllib2.urlopen(corpus_URL) # it's a file like object and works just like a file\n",
" for file_URL in url_list_text_data: # files are iterable\n",
" yield file_URL.rstrip()"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 1
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Use these commands to read in a list of filenames (samples) in a directory, replacing the URL with a URL to a list of URLs (one per line) indicating where the audio / drum samples are stored."
"kick_file_list = [audio_file_URL for audio_file_URL in process_corpus(kicks_URL)]"
],
"language": "python",
"metadata": {},
"outputs": [],
"prompt_number": 3
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"To access the filenames contained in the array, use the square brackets [ ] to get to the element that you want to access. For example, to access the text URL file name of the first file in the list, you would type:"
"When we feature extract a sample collection, we need to sequentially access audio files, segment them (or not), and feature extract them. Loading a lot of audio files into memory is not always a feasible or desirable operation, so you will create a loop which loads an audio file, feature extracts it, and closes the audio file. Note that the only information that we retain in memory are the features that are extracted.\n",
"\n",
"Create a loop which reads in an audio file, extracts the zero crossing rate, and some spectral statistics. You can use the \"in\" operator to retrieve each audio file URL from process_corpus(), as used above. The feature information for each audio file (the \"feature vector\") should be stored as a feature array, with columns being the features and rows for each file. For example:\n",
"\n",
" featuresSnare =\n",
"\n",
" 0.5730 1.9183 2.9713 0.0004 0.0002\n",
" 0.4750 1.4834 2.4463 0.0004 0.0012\n",
" 0.5900 2.2857 3.1788 0.0003 0.0041\n",
" 0.5090 1.6622 2.6369 0.0004 0.0051\n",
" 0.4860 1.4758 2.2085 0.0004 0.0021\n",
" 0.6060 2.2119 3.2798 0.0004 0.0651\n",
" 0.4990 2.0607 2.7654 0.0004 0.0721\n",
" 0.6360 2.3153 3.0256 0.0003 0.0221\n",
" 0.5490 2.0137 3.0342 0.0004 0.0016\n",
" 0.5900 2.2857 3.1788 0.0003 0.0012\n",
" \n",
" Within your loop, here's a reminder how to read in your wav files, using an array of audio file URLs:"
"4. First, extract all of the feature data for the kick drums and store it in a feature array. (For my example, above, I'd put it in \"featuresKick\")\n",
"\n",
"5. Next, extract all of the feature data for the snares, storing them in a different array. \n",
"Again, the kick and snare features should be separated in two different arrays!\n",
