What is the error " TypeError: 'int' object is not callable"?

Question

Substitute three numpy for the audio and combine them to get the max-min average. I am getting an error with this, what should I do?

import torch
import torchaudio
import torchaudio.transforms as T
import os
import requests
import librosa
import matplotlib.pyplot as plt
# 音声の保存
_SAMPLE_DIR = "_sample_data"
SAMPLE_WAV_URL = "https://pytorch-tutorial-assets.s3.amazonaws.com/VOiCES_devkit/source-16k/train/sp0307/Lab41-SRI-VOiCES-src-sp0307-ch127535-sg0042.wav"
SAMPLE_WAV_PATH = os.path.join(_SAMPLE_DIR, "speech.wav")
def plot_spectrogram(spec, title=None, ylabel="freq_bin", aspect="auto", xmax=None):
    fig, axs = plt.subplots(1, 1)
    axs.set_title(title or "Spectrogram (db)")
    axs.set_ylabel(ylabel)
    axs.set_xlabel("frame")
    im = axs.imshow(librosa.power_to_db(spec), origin="lower", aspect=aspect)
    if xmax:
        axs.set_xlim((0, xmax))
    fig.colorbar(im, ax=axs)
    plt.show(block=False)
def synthesis(sigList):
    maxLength = 0
    tmpLength = 0
    tmpArray = []
    #最大長の音声を探索する
    for i, data in enumerate(sigList):
        if len(data) > tmpLength:
            maxLength = len(data)
            tmpLength = len(data)
            index = i
    #最大長の音声の長さの０埋め配列を定義
    sig = np.zeros(maxLength)

    for i in sigList:
        tmp = i.tolist()    #numpy→list
        #全ての音声を最大長の音声に合わせて０埋めする
        for data in range(maxLength - len(i)):
            tmp.append(0)
        tmpArray.append(tmp)
    #配列３つを合成する
    sig = np.array(tmpArray[0]) + np.array(tmpArray[1]) + np.array(tmpArray[2])
    return sig
def min_max(x, axis=None):
    min = x.min(axis=axis, keepdims=True)
    max = x.max(axis=axis, keepdims=True)
    try:
        z = (x - min) / (max - min)
    except ZeroDivisionError:
        z = (x - min) / min
    return z
waveform, sample_rate = torchaudio.load(filepath=SAMPLE_WAV_URL)

n_fft = 1024
win_length = None
hop_length = 512
window_fn = torch.hann_window
waveforms = waveform.numpy()
k = waveforms
for i in range(2):
  waveforms = np.concatenate([waveforms,k],0)
spectrogram = T.Spectrogram(
    n_fft=n_fft,
    win_length=win_length,
    hop_length=hop_length,
    window_fn=window_fn,
    power=2.0,
)

sig  = min_max(synthesis(waveforms))

spec = spectrogram(sig)
plot_spectrogram(spec[0], title='torchaudio')

spec = spectrogram(sig) This is the line where the error occurs. Detailed error is TypeError

Traceback (most recent call last)
<ipython-input-44-a0a6c4ba7770> in <module>
      70 sig = min_max(synthesis(waveforms))
      71
---> 72 spec = spectrogram(sig)
      73 plot_spectrogram(spec[0], title='torchaudio')

2 frames
/usr/local/lib/python3.7/dist-packages/torchaudio/functional/functional.py in spectrogram(waveform, pad, window, n_fft, hop_length, win_length, power, normalized, center, pad_mode, onesided, return_complex)
     106
     107# pack batch
--> 108 shape = waveform.size()
     109 waveform = waveform.reshape(-1, shape[-1])
     110

TypeError: 'int' object is not callable

Answer 1

According to the docs for Torchaudio Spectrogram, the parameter that's passed to its return value (spectrogram() in your code) needs to be a PyTorch Tensor. In your code, you're giving it a Numpy array instead, because that's what your function synthesis() returns.

You can convert a Numpy ndarray into a Tensor with torch.from_numpy. For example:

spec = spectrogram(torch.from_numpy(sig))