如何用Python将(频率,振幅)数据转换为音频？已试IPython.display.audio未成功

Absolutely! Turning your vibrating string's frequency-amplitude data into playable audio is totally feasible in Python. Let’s walk through the exact steps to get this working, including fixing the issues that might have broken your initial IPython.display.audio attempt.

Step 1: Understand the Core Idea #

Audio players work with time-domain waveforms—not raw frequency-amplitude pairs. So we need to convert your frequency data back into a continuous signal by summing up sine waves for each (frequency, amplitude) pair (since Fourier analysis tells us any periodic signal is a sum of sinusoids).

Step 2: Full Implementation Code #

Let’s use numpy to build the waveform and IPython.display.Audio to play it. Here’s a complete, runnable example:

import numpy as np
from IPython.display import Audio

# Replace this with your actual (frequency, amplitude) data
freq_amp_pairs = [(220, 0.5), (440, 0.8), (660, 0.3), (880, 0.2)]

# Configure audio parameters
sample_rate = 44100  # Standard CD-quality sampling rate (Hz)
duration = 3  # Length of your audio in seconds

# Generate a time axis for the waveform
time_axis = np.linspace(0, duration, int(sample_rate * duration), endpoint=False)

# Initialize an empty audio signal
audio_waveform = np.zeros_like(time_axis)

# Add each frequency component to the waveform
for freq, amp in freq_amp_pairs:
    # Sine wave formula: amplitude * sin(2π * frequency * time)
    audio_waveform += amp * np.sin(2 * np.pi * freq * time_axis)

# Critical: Normalize the signal to avoid clipping/distortion
# We scale the waveform so its peak value is 1 (the max for audio)
audio_waveform = audio_waveform / np.max(np.abs(audio_waveform))

# Play the audio directly in IPython/Jupyter
Audio(audio_waveform, rate=sample_rate)

Why Your Initial Attempt Might Have Failed #

Most common issues:

• No normalization: If your signal’s amplitude exceeded [-1, 1], Audio would silently fail or produce distorted sound. The code above fixes this with the normalization step.
• Incorrect frequency values: Double-check that your frequencies are in Hz (not radians/s or another unit) and don’t include a 0Hz (DC) component (which would add a silent offset).
• Missing time-domain conversion: If you tried passing raw frequency-amplitude pairs directly to Audio, it wouldn’t know how to interpret them—you need to build the waveform first.

If You’re Working Directly with FFT Results #

If you have the full Fourier transform output (complex numbers) instead of just (freq, amp) pairs, use inverse FFT to get the waveform:

import numpy as np
from IPython.display import Audio

# Replace fft_result with your actual Fourier transform output
fft_result = np.fft.fft(your_time_domain_signal)  # Example FFT (replace with your data)

# Use inverse FFT to get back the time-domain signal
audio_waveform = np.fft.irfft(fft_result)

# Normalize and play
audio_waveform = audio_waveform / np.max(np.abs(audio_waveform))
Audio(audio_waveform, rate=sample_rate)

Save the Audio to a File #

If you want to export the sound as a WAV file (for use outside IPython), use scipy.io.wavfile:

from scipy.io.wavfile import write

# Convert to 16-bit integer (standard WAV format)
audio_int = np.int16(audio_waveform * 32767)  # 32767 is the max value for 16-bit ints

# Save the file
write("vibrating_string_sound.wav", sample_rate, audio_int)

内容的提问来源于stack exchange，提问作者Nogniks