{"cells":[{"metadata":{},"cell_type":"markdown","source":"
\n
\n
\n
\n

Free Sound Cloud Audio Classification Challange

\"\"

\n

Project Model

Input file : Audio Wave file
\nOutput file : Label\n\"\"

\n\n
\n
\n
"},{"metadata":{"toc":true},"cell_type":"markdown","source":"

Table of Contents

\n
"},{"metadata":{},"cell_type":"markdown","source":"## Loading an audio file"},{"metadata":{"trusted":true},"cell_type":"code","source":"import warnings\nwarnings.filterwarnings('ignore')","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"import librosa\naudio_path = '../input/train_curated/0006ae4e.wav'\nx , sr = librosa.load(audio_path)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Playing Audio\nUsing IPython.display.Audio, to play the audio"},{"metadata":{"trusted":true},"cell_type":"code","source":"import IPython.display as ipd\nipd.Audio(audio_path)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"You can even use an mp3 or a WMA format for the audio example."},{"metadata":{},"cell_type":"markdown","source":"## Visualizing Audio\n\n### Waveform\nWe can plot the audio array using librosa.display.waveplot:\n"},{"metadata":{"trusted":true},"cell_type":"code","source":"%matplotlib inline\nimport sklearn\nimport matplotlib.pyplot as plt\nimport librosa.display\n\nplt.figure(figsize=(20, 5))\nlibrosa.display.waveplot(x, sr=sr)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"Here, we have the plot the amplitude envelope of a waveform."},{"metadata":{},"cell_type":"markdown","source":"### Spectrogram\nWe can also display a spectrogram using librosa.display.specshow."},{"metadata":{"trusted":true},"cell_type":"code","source":"X = librosa.stft(x)\nXdb = librosa.amplitude_to_db(abs(X))\nplt.figure(figsize=(20, 5))\nlibrosa.display.specshow(Xdb, sr=sr, x_axis='time', y_axis='hz')\nplt.colorbar()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"### Log Frequency axis"},{"metadata":{"trusted":true},"cell_type":"code","source":"librosa.display.specshow(Xdb, sr=sr, x_axis='time', y_axis='log')\nplt.colorbar()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Creating an audio signal\nLet us now create an audio signal at 220Hz. We know an audio signal is a numpy array, so we shall create one and pass it on to the audio function.\n"},{"metadata":{"trusted":true},"cell_type":"code","source":"import numpy as np\nsr = 22050 # sample rate\nT = 5.0 # seconds\nt = np.linspace(0, T, int(T*sr), endpoint=False) # time variable\nx = 0.5*np.sin(2*np.pi*220*t)# pure sine wave at 220 Hz\n","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Playing the sound\n"},{"metadata":{"trusted":true},"cell_type":"code","source":"ipd.Audio(x, rate=sr) # load a NumPy array","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Saving the signal\n"},{"metadata":{"trusted":true},"cell_type":"code","source":"librosa.output.write_wav('../tone_440.wav', x, sr) # writing wave file in tone440.wav format","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Feature Extraction"},{"metadata":{"trusted":true},"cell_type":"code","source":"x, sr = librosa.load('../input/train_noisy/000b6cfb.wav')\nipd.Audio(x, rate=sr)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"#Plot the signal:\nplt.figure(figsize=(20, 5))\nlibrosa.display.waveplot(x, sr=sr)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## 1. Zero Crossing Rate"},{"metadata":{"trusted":true},"cell_type":"code","source":"# Zooming in\nn0 = 9000\nn1 = 9100\nplt.figure(figsize=(20, 5))\nplt.plot(x[n0:n1])\nplt.grid()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"I count 6 zero crossings. Let's compute the zero crossings using librosa."},{"metadata":{"trusted":true},"cell_type":"code","source":"zero_crossings = librosa.zero_crossings(x[n0:n1], pad=False)\nzero_crossings.shape","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"print(sum(zero_crossings))","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## 2.Spectral Centroid"},{"metadata":{"trusted":true},"cell_type":"code","source":"spectral_centroids = librosa.feature.spectral_centroid(x, sr=sr)[0]\nspectral_centroids.shape","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# Computing the time variable for visualization\nplt.figure(figsize=(20,5))\nframes = range(len(spectral_centroids))\nt = librosa.frames_to_time(frames)\n\n# Normalising the spectral centroid for visualisation\ndef normalize(x, axis=0):\n return sklearn.preprocessing.minmax_scale(x, axis=axis)\n\n#Plotting the Spectral Centroid along the waveform\nlibrosa.display.waveplot(x, sr=sr, alpha=0.4)\nplt.plot(t, normalize(spectral_centroids), color='r')","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## 3.Spectral Rolloff "},{"metadata":{"trusted":true},"cell_type":"code","source":"plt.figure(figsize=(20,5))\nspectral_rolloff = librosa.feature.spectral_rolloff(x+0.01, sr=sr)[0]\nlibrosa.display.waveplot(x, sr=sr, alpha=0.4)\nplt.plot(t, normalize(spectral_rolloff), color='r')\nplt.grid()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## 4.MFCC"},{"metadata":{"trusted":true},"cell_type":"code","source":"plt.figure(figsize=(20,5))\nx, fs = librosa.load('../input/train_curated/0006ae4e.wav')\nlibrosa.display.waveplot(x, sr=sr)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# MFCC\nplt.figure(figsize=(20,5))\nmfccs = librosa.feature.mfcc(x, sr=sr)\nprint(mfccs.shape)\n\nlibrosa.display.specshow(mfccs, sr=sr, x_axis='time')","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Feature Scaling\nLet's scale the MFCCs such that each coefficient dimension has zero mean and unit variance:"},{"metadata":{"trusted":true},"cell_type":"code","source":"mfccs = sklearn.preprocessing.scale(mfccs, axis=1)\nprint(mfccs.mean(axis=1))\nprint(mfccs.var(axis=1))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"plt.figure(figsize=(20,8))\nlibrosa.display.specshow(mfccs, sr=sr, x_axis='time')","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Chroma Frequencies"},{"metadata":{"trusted":true},"cell_type":"code","source":"# Loadign the file\nx, sr = librosa.load('../input/train_curated/0006ae4e.wav')\nipd.Audio(x, rate=sr)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"hop_length = 512\nchromagram = librosa.feature.chroma_stft(x, sr=sr, hop_length=hop_length)\nplt.figure(figsize=(15, 5))\nlibrosa.display.specshow(chromagram, x_axis='time', y_axis='chroma', hop_length=hop_length, cmap='coolwarm')","execution_count":null,"outputs":[]}],"metadata":{"gist":{"data":{"description":"Desktop/Audio Analysis in Python.ipynb","public":false},"id":""},"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":"3.6.0"},"toc":{"base_numbering":1,"nav_menu":{},"number_sections":true,"sideBar":true,"skip_h1_title":false,"title_cell":"Table of Contents","title_sidebar":"Contents","toc_cell":true,"toc_position":{"height":"calc(100% - 180px)","left":"10px","top":"150px","width":"259.901px"},"toc_section_display":true,"toc_window_display":false}},"nbformat":4,"nbformat_minor":1}