Hey there, sorry this has been eating up so much of your time—let’s walk through some common checks to pinpoint where things might be going wrong with your Librosa audio loading and plotting code.

First, let’s start with the basics around the y array (the audio time series) since that’s the lead suspect you mentioned:

• Verify y type and shape for each audio file: Librosa’s librosa.load() should return a 1D numpy array for mono audio (the default behavior). Run this quick check for each file to confirm:

  import librosa
  import numpy as np
  
  for audio_path in ["file1.wav", "file2.wav", "file3.wav"]:
      y, sr = librosa.load(audio_path)
      print(f"File: {audio_path}")
      print(f"Type of y: {type(y)}")
      print(f"Shape of y: {y.shape}")
      print(f"Is y a valid numpy array? {isinstance(y, np.ndarray)}")
      print("---")
  

  If any of these show y isn’t a numpy array (e.g., a Python list) or has an unexpected shape (like 2D for multi-channel audio without handling), that’s likely the culprit. For multi-channel files, add y = librosa.to_mono(y) right after loading to convert to single-channel.

Next, check these other common pain points:

• Ensure consistent sampling rates: If your three audio files have different native sampling rates, plotting them without normalizing might cause unexpected errors or misalignment. Force a fixed sampling rate when loading:

  y, sr = librosa.load(audio_path, sr=22050)  # 22050 is Librosa's default, adjust as needed
  

• Validate audio integrity: Corrupted audio files can lead to malformed y arrays. Use Librosa’s built-in check to confirm each loaded audio is valid:

  if not librosa.util.valid_audio(y):
      print(f"Invalid audio data in {audio_path}")
  

• Double-check your plotting code: If the loading checks out, the issue might be in how you’re plotting. For example, if you’re looping through the three audio files and reusing the same plot axis without clearing it, or trying to plot a 2D array directly. A simple plotting example for reference:

  import matplotlib.pyplot as plt
  
  fig, axes = plt.subplots(3, 1, figsize=(10, 8))
  for idx, audio_path in enumerate(["file1.wav", "file2.wav", "file3.wav"]):
      y, sr = librosa.load(audio_path, sr=22050)
      librosa.display.waveshow(y, sr=sr, ax=axes[idx])
      axes[idx].set_title(f"Audio File {idx+1}")
  plt.tight_layout()
  plt.show()
  

If none of these fix the issue, sharing the full error traceback and your exact code would help narrow things down even more—error messages often have specific clues about what’s mismatched or invalid.

内容的提问来源于stack exchange，提问作者user8716125