AudioHandling

ipynb4 kBPython · 5 cellsSolution of TP 2: Audio and video handling

Audio Handling (Using SciPy)

For handling standard uncompressed audio (like WAV files), the scipy.io.wavfile module is simple and effective for a conceptual course.

In [ ]
import numpy as np
from scipy.io import wavfile
import matplotlib.pyplot as plt

1. Reading and Seeing Audio Data

Audio data is a sequence of samples over time.

In [ ]
def analyze_audio(audio_path):
    # 1. Read the audio file
    # rate: Sampling rate (samples/second, e.g., 44100 Hz)
    # data: The actual audio samples (a NumPy array)
    try:
        rate, data = wavfile.read(audio_path)
    except FileNotFoundError:
        print("Error: Could not find audio file.")
        return

    print(f"--- Audio Metadata ---")
    print(f"Sampling Rate (Hz): {rate}")

    # Check if stereo (2 channels) or mono (1 channel)
    if data.ndim > 1:
        channels = data.shape[1]
        print(f"Number of Channels: {channels}")
        # Use the first channel for plotting if stereo
        audio_data = data[:, 0]
    else:
        channels = 1
        print(f"Number of Channels: {channels}")
        audio_data = data

    # 2. Inspect the data (e.g., the first 10 samples)
    print(f"Total number of samples: {len(audio_data)}")
    print(
        f"Data type (Bit Depth): {audio_data.dtype} (e.g., int16 means 16-bit resolution)"
    )
    print(f"First 10 samples: {audio_data[:10]}")

    # 3. Visualize the signal (in the Spatial/Time Domain)
    # Create a time vector
    time = np.linspace(0.0, len(audio_data) / rate, len(audio_data))

    plt.figure(figsize=(10, 4))
    plt.plot(time, audio_data)
    plt.title(f"Audio Signal (Time Domain - {channels} Channel(s))")
    plt.xlabel("Time (seconds)")
    plt.ylabel("Amplitude (Sample Value)")
    plt.show()


# Example Usage: analyze_audio('path/to/your/audio.wav')

Key Audio Concepts:

  • Sampling Rate (rate): How many samples (digital amplitude values) are taken per second.
  • Amplitude/Bit Depth (data.dtype): The resolution of each sample (e.g., 8-bit, 16-bit). This relates directly to the maximum and minimum sample values (e.g., ±32767 for 16-bit signed integers).
  • Time Domain: The plot shows amplitude (y-axis) vs. time (x-axis), representing the sound wave.

In TP 2: Audio and video handling