vision_features_bokeh3.ipynb

{"cells":[{"cell_type":"code","execution_count":1,"id":"01764acb-cd19-48f8-8c28-e59b5d60a4f2","metadata":{"id":"01764acb-cd19-48f8-8c28-e59b5d60a4f2","executionInfo":{"status":"ok","timestamp":1727050701614,"user_tz":-120,"elapsed":3965,"user":{"displayName":"Egor P","userId":"11648005462717752555"}}},"outputs":[],"source":["# Import necessary libraries\n","import os\n","import warnings\n","import requests\n","import librosa\n","import numpy as np\n","import pandas as pd\n","from bokeh.plotting import figure, show, output_notebook\n","from bokeh.layouts import column\n","from bokeh.models import Span, LinearColorMapper, ColorBar, FixedTicker\n","\n","# Initialize Bokeh for notebook output\n","output_notebook()\n","\n","\n"]},{"cell_type":"code","execution_count":2,"id":"a9d1b29c-7c55-40a5-a43a-2e8e4a56a4ed","metadata":{"id":"a9d1b29c-7c55-40a5-a43a-2e8e4a56a4ed","executionInfo":{"status":"ok","timestamp":1727050701614,"user_tz":-120,"elapsed":4,"user":{"displayName":"Egor P","userId":"11648005462717752555"}}},"outputs":[],"source":["# Define utility functions with docstrings\n","def download_audio(audio_url: str, local_path: str):\n","    \"\"\"\n","    Download audio file from a URL to a local path.\n","\n","    Parameters:\n","    - audio_url: URL of the audio file to download.\n","    - local_path: Local path to save the downloaded audio file.\n","    \"\"\"\n","    if not os.path.exists(local_path):\n","        print(f\"Downloading audio from {audio_url}...\")\n","        response = requests.get(audio_url)\n","        response.raise_for_status()  # Raise an error for bad status\n","        with open(local_path, 'wb') as f:\n","            f.write(response.content)\n","        print(f\"Downloaded audio to {local_path}.\")\n","    else:\n","        print(f\"Audio file {local_path} already exists. Skipping download.\")\n","\n","def extract_features(y_segment: np.ndarray, sr: int, hop_length: int):\n","    \"\"\"\n","    Extract various audio features from a segment.\n","\n","    Parameters:\n","    - y_segment: Audio time series.\n","    - sr: Sampling rate.\n","    - hop_length: Hop length for feature calculation.\n","\n","    Returns:\n","    - features: Dictionary containing extracted features.\n","    \"\"\"\n","    features = {}\n","    features['amplitude_envelope'] = librosa.onset.onset_strength(y=y_segment, sr=sr, hop_length=hop_length)\n","    S = librosa.feature.melspectrogram(y=y_segment, sr=sr, hop_length=hop_length)\n","    features['melspectrogram'] = S\n","    features['spectral_flux'] = np.diff(S, axis=1)\n","    features['zero_crossing_rate'] = librosa.feature.zero_crossing_rate(y_segment, hop_length=hop_length)[0]\n","    features['spectral_centroid'] = librosa.feature.spectral_centroid(y=y_segment, sr=sr, hop_length=hop_length)[0]\n","    return features\n","\n","def create_time_axis(num_frames: int, sr: int, hop_length: int, start_time: float):\n","    \"\"\"\n","    Create a time axis for plotting.\n","\n","    Parameters:\n","    - num_frames: Number of frames.\n","    - sr: Sampling rate.\n","    - hop_length: Hop length.\n","    - start_time: Start time in seconds.\n","\n","    Returns:\n","    - Time axis array.\n","    \"\"\"\n","    return librosa.frames_to_time(np.arange(num_frames), sr=sr, hop_length=hop_length) + start_time\n","\n","def plot_stft_spectrogram(y_segment: np.ndarray, sr: int, n_fft: int, hop_length_stft: int, y_min: int, y_max: int, start_time: float, title: str, width=400, height=300, interp_factor=0.5):\n","    \"\"\"\n","    Create a Bokeh figure for the STFT spectrogram.\n","\n","    Parameters:\n","    - y_segment: Audio segment.\n","    - sr: Sampling rate.\n","    - n_fft: FFT window size.\n","    - hop_length_stft: Hop length for STFT.\n","    - y_min: Minimum frequency to display.\n","    - y_max: Maximum frequency to display.\n","    - start_time: Start time of the segment.\n","    - title: Plot title.\n","    - width: Plot width.\n","    - height: Plot height.\n","    - interp_factor: Interpolation factor between linear and log frequencies.\n","\n","    Returns:\n","    - Bokeh figure.\n","    \"\"\"\n","    stft = librosa.stft(y=y_segment, n_fft=n_fft, hop_length=hop_length_stft)\n","    stft_db_full = librosa.amplitude_to_db(np.abs(stft), ref=np.max)\n","\n","    freqs_full = np.linspace(0, sr / 2, num=stft_db_full.shape[0])\n","    freq_indices = np.where((freqs_full >= y_min) & (freqs_full <= y_max))[0]\n","\n","    stft_db = stft_db_full[freq_indices, :]\n","    freqs = freqs_full[freq_indices]\n","\n","    # Interpolation between linear and log frequencies\n","    linear_freqs = np.linspace(y_min, y_max, num=len(freq_indices))\n","    log_freqs = np.logspace(np.log10(max(y_min, 1)), np.log10(y_max), num=len(freq_indices))\n","    interp_freqs = (1 - interp_factor) * linear_freqs + interp_factor * log_freqs\n","\n","    num_ticks = min(10, len(interp_freqs))\n","    tick_indices = np.linspace(0, len(interp_freqs) - 1, num=num_ticks).astype(int)\n","    tick_locations = interp_freqs[tick_indices]\n","    tick_labels = [f\"{int(freq)} Hz\" for freq in tick_locations]\n","\n","    p = figure(title=title, x_axis_label='Time (s)', y_axis_label='Frequency (Hz)',\n","               width=width, height=height, y_range=(y_min, y_max))\n","\n","    color_mapper = LinearColorMapper(palette=\"Viridis256\", low=np.min(stft_db), high=np.max(stft_db))\n","\n","    times_stft = librosa.frames_to_time(np.arange(stft_db.shape[1]), sr=sr, hop_length=hop_length_stft) + start_time\n","\n","    p.image(image=[stft_db],\n","            x=times_stft[0],\n","            y=y_min,\n","            dw=times_stft[-1] - times_stft[0],\n","            dh=y_max - y_min,\n","            color_mapper=color_mapper)\n","\n","    p.yaxis.ticker = FixedTicker(ticks=tick_locations.tolist())\n","    p.yaxis.major_label_overrides = {freq: label for freq, label in zip(tick_locations, tick_labels)}\n","\n","    color_bar = ColorBar(\n","        color_mapper=color_mapper,\n","        label_standoff=12,\n","        location=(0, 0),\n","        title='dB'\n","    )\n","    p.add_layout(color_bar, 'right')\n","\n","    return p\n","\n","def add_custom_lines(plot, custom_intervals, duration):\n","    \"\"\"\n","    Add custom interval lines to a plot.\n","\n","    Parameters:\n","    - plot: Bokeh figure to add lines to.\n","    - custom_intervals: List of dictionaries defining intervals.\n","    - duration: Duration of the audio segment.\n","    \"\"\"\n","    for interval_set in custom_intervals:\n","        start = interval_set['start_time']\n","        step = interval_set['interval']\n","        color = interval_set.get('color', 'red')\n","        custom_time = start\n","        while custom_time < duration:\n","            span = Span(location=custom_time, dimension='height', line_color=color,\n","                        line_dash='dotted', line_width=2)\n","            plot.add_layout(span)\n","            custom_time += step\n"]},{"cell_type":"code","execution_count":3,"id":"199fc01f-3309-4579-8028-f01ab41c5cf8","metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"199fc01f-3309-4579-8028-f01ab41c5cf8","executionInfo":{"status":"ok","timestamp":1727050739811,"user_tz":-120,"elapsed":38201,"user":{"displayName":"Egor P","userId":"11648005462717752555"}},"outputId":"211619cd-b3da-42f3-bbd0-7d64ace45d53"},"outputs":[{"output_type":"stream","name":"stdout","text":["Downloading audio from https://github.com/egorpol/beat_it/raw/refs/heads/main/examples/vision_spectral_mono.aif...\n","Downloaded audio to vision_spectral_mono.aif.\n","The audio file is 73.33 seconds long.\n"]}],"source":["# Set parameters for audio processing\n","audio_url = 'https://github.com/egorpol/beat_it/raw/refs/heads/main/examples/vision_spectral_mono.aif'\n","local_audio_path = 'vision_spectral_mono.aif'\n","\n","# Download and load the audio file\n","download_audio(audio_url, local_audio_path)\n","\n","# Load audio using librosa\n","y, sr = librosa.load(local_audio_path)\n","duration = len(y) / sr\n","print(f\"The audio file is {duration:.2f} seconds long.\")\n"]},{"cell_type":"code","execution_count":4,"id":"de950460-01f7-4ac3-8122-46fa06ed82cf","metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"de950460-01f7-4ac3-8122-46fa06ed82cf","executionInfo":{"status":"ok","timestamp":1727050739811,"user_tz":-120,"elapsed":6,"user":{"displayName":"Egor P","userId":"11648005462717752555"}},"outputId":"7eb09e72-ed44-4364-c1c1-75cc1e5f392b"},"outputs":[{"output_type":"stream","name":"stdout","text":["Grouped Segments (Start - End in seconds):\n","Group 1: 0.103 - 14.765 s\n","Group 2: 14.765 - 29.428 s\n","Group 3: 29.428 - 44.090 s\n","Group 4: 44.090 - 58.753 s\n"]}],"source":["# STFT parameters\n","interp_factor = 0         # 0 for linear frequency scale, 1 for logarithmic\n","window_size = 1024\n","overlap = 0.5\n","hop_length_stft = int(window_size * (1 - overlap))\n","n_fft = window_size\n","y_min = 0                 # Minimum frequency to display in Hz\n","y_max = 10000             # Maximum frequency to display in Hz\n","\n","# Parameters for grouped analysis\n","initial_start_time = 0.103       # Starting time of the first interval in seconds\n","interval_duration = 0.4582       # Duration of each custom interval in seconds\n","intervals_per_group = 32         # Number of intervals per group\n","total_groups = 4                 # Number of groups to analyze\n","\n","# Compute start_times and end_times for each group\n","start_times_grouped = []\n","end_times_grouped = []\n","\n","for group in range(total_groups):\n","    start_time = initial_start_time + group * intervals_per_group * interval_duration\n","    end_time = start_time + intervals_per_group * interval_duration\n","    # Ensure end_time does not exceed audio duration\n","    if end_time > duration:\n","        end_time = duration\n","    start_times_grouped.append(start_time)\n","    end_times_grouped.append(end_time)\n","\n","print(\"Grouped Segments (Start - End in seconds):\")\n","for idx, (s, e) in enumerate(zip(start_times_grouped, end_times_grouped), 1):\n","    print(f\"Group {idx}: {s:.3f} - {e:.3f} s\")\n"]},{"cell_type":"code","execution_count":5,"id":"e30f8018-1d51-4dbd-8e42-db59ddd5dc63","metadata":{"id":"e30f8018-1d51-4dbd-8e42-db59ddd5dc63","executionInfo":{"status":"ok","timestamp":1727050739811,"user_tz":-120,"elapsed":4,"user":{"displayName":"Egor P","userId":"11648005462717752555"}}},"outputs":[],"source":["# Define custom interval lines (e.g., beat intervals)\n","custom_intervals = [\n","    {'start_time': 0.103, 'interval': 0.4582, 'color': 'orange'},\n","    # Add more custom intervals here if needed\n","]\n"]},{"cell_type":"code","execution_count":6,"id":"2173e3c1-bc6e-4eb7-86d4-2f7ca563f7d9","metadata":{"colab":{"base_uri":"https://localhost:8080/","height":1000,"output_embedded_package_id":"1lE8y8evrfX4lGLejmWOAoPZcr4ycy24W"},"id":"2173e3c1-bc6e-4eb7-86d4-2f7ca563f7d9","executionInfo":{"status":"ok","timestamp":1727050745083,"user_tz":-120,"elapsed":5275,"user":{"displayName":"Egor P","userId":"11648005462717752555"}},"outputId":"b9ba7bda-6bdc-42fc-d3e7-6664aa989e0a"},"outputs":[{"output_type":"display_data","data":{"text/plain":"Output hidden; open in https://colab.research.google.com to view."},"metadata":{}}],"source":["# Process each grouped segment and plot the STFT spectrogram\n","group_layouts = []  # List to hold layouts for each group\n","\n","for idx, (start_time, end_time) in enumerate(zip(start_times_grouped, end_times_grouped), 1):\n","    print(f\"\\nProcessing Group {idx}: {start_time:.3f} - {end_time:.3f} seconds\")\n","\n","    # Extract the audio segment for the current group\n","    start_sample = int(start_time * sr)\n","    end_sample = int(end_time * sr)\n","    y_segment = y[start_sample:end_sample]\n","\n","    # Plot STFT spectrogram\n","    p_stft = plot_stft_spectrogram(\n","        y_segment, sr, n_fft, hop_length_stft, y_min, y_max,\n","        start_time, f\"STFT Spectrogram (Group {idx})\", width=1200, height=600, interp_factor=interp_factor\n","    )\n","\n","    # Add custom interval lines to the plot\n","    add_custom_lines(p_stft, custom_intervals, end_time)\n","\n","    # Add the plot to the list of group layouts\n","    group_layouts.append(p_stft)\n","\n","# Arrange all group layouts vertically and display\n","final_layout = column(*group_layouts, sizing_mode='stretch_both')\n","show(final_layout)\n"]},{"cell_type":"code","execution_count":7,"id":"8464f4eb-7849-4f07-a7c4-280185a77269","metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"8464f4eb-7849-4f07-a7c4-280185a77269","executionInfo":{"status":"ok","timestamp":1727050745083,"user_tz":-120,"elapsed":8,"user":{"displayName":"Egor P","userId":"11648005462717752555"}},"outputId":"ad158167-7383-4544-8996-6c429f05a0a5"},"outputs":[{"output_type":"stream","name":"stdout","text":["Pairwise Cosine Similarities between Groups:\n","[[1.         0.95164469 0.94994353 0.94551739]\n"," [0.95164469 1.         0.95586015 0.95153916]\n"," [0.94994353 0.95586015 1.         0.95594179]\n"," [0.94551739 0.95153916 0.95594179 1.        ]]\n"]}],"source":["import numpy as np\n","from scipy.spatial.distance import cosine\n","\n","# List to store flattened STFT spectrograms for each group\n","stft_vectors = []\n","\n","for idx, (start_time, end_time) in enumerate(zip(start_times_grouped, end_times_grouped), 1):\n","    # Extract the audio segment\n","    start_sample = int(start_time * sr)\n","    end_sample = int(end_time * sr)\n","    y_segment = y[start_sample:end_sample]\n","\n","    # Compute STFT\n","    stft = librosa.stft(y=y_segment, n_fft=n_fft, hop_length=hop_length_stft)\n","    stft_magnitude = np.abs(stft)\n","\n","    # Flatten the magnitude spectrogram\n","    stft_vector = stft_magnitude.flatten()\n","\n","    # Normalize the vector\n","    stft_vector /= np.linalg.norm(stft_vector)\n","\n","    stft_vectors.append(stft_vector)\n","\n","# Initialize similarity matrix\n","num_groups = len(stft_vectors)\n","similarity_matrix = np.zeros((num_groups, num_groups))\n","\n","# Compute pairwise cosine similarity\n","for i in range(num_groups):\n","    for j in range(num_groups):\n","        similarity = 1 - cosine(stft_vectors[i], stft_vectors[j])\n","        similarity_matrix[i, j] = similarity\n","\n","# Display the similarity matrix\n","print(\"Pairwise Cosine Similarities between Groups:\")\n","print(similarity_matrix)\n"]},{"cell_type":"code","execution_count":8,"id":"dd68c20d-e364-4a44-a87f-7b8a22f99ed0","metadata":{"colab":{"base_uri":"https://localhost:8080/","height":545},"id":"dd68c20d-e364-4a44-a87f-7b8a22f99ed0","executionInfo":{"status":"ok","timestamp":1727050745083,"user_tz":-120,"elapsed":7,"user":{"displayName":"Egor P","userId":"11648005462717752555"}},"outputId":"62929201-47fb-4bed-d1ba-52972c2823fd"},"outputs":[{"output_type":"display_data","data":{"text/plain":["<Figure size 800x600 with 2 Axes>"],"image/png":"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\n"},"metadata":{}}],"source":["import seaborn as sns\n","import matplotlib.pyplot as plt\n","\n","# Create a heatmap\n","group_labels = [f'Group {i+1}' for i in range(num_groups)]\n","plt.figure(figsize=(8, 6))\n","sns.heatmap(similarity_matrix, annot=True, fmt=\".2f\", cmap='viridis',\n","            xticklabels=group_labels, yticklabels=group_labels)\n","plt.title('Cosine Similarity between Groups')\n","plt.show()\n"]},{"cell_type":"code","execution_count":8,"id":"c6c955fb-aa53-4479-ab6b-4a7c021862c5","metadata":{"id":"c6c955fb-aa53-4479-ab6b-4a7c021862c5","executionInfo":{"status":"ok","timestamp":1727050745084,"user_tz":-120,"elapsed":7,"user":{"displayName":"Egor P","userId":"11648005462717752555"}}},"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":"3.12.4"},"colab":{"provenance":[]}},"nbformat":4,"nbformat_minor":5}