Question

我们已经编程了一种设备（由ra berry pi 3组成），可以坐在一个房间里，并根据现有的GMM对房间进行表征。我们的代码可以成功运行，直到在第17或19次迭代中引发错误为止。我们已经缩小了代码在哪里出现问题的范围，但是我们不确定为什么。

我们是初学者，所以为怪异的格式道歉...这对我们来说都是新的。

在此先感谢您提供的任何帮助或见解！

sz

我们认为问题发生在这一行：

from sklearn import preprocessing
import python_speech_features as mfcc

def calculate_delta(array):
    """Calculate and returns the delta of given feature vector matrix"""

    rows,cols = array.shape
    deltas = np.zeros((rows,20))
    N = 2
    for i in range(rows):
        index = []
        j = 1
        while j <= N:
            if i-j < 0:
                first = 0
            else:
                first = i-j
            if i+j > rows -1:
                second = rows -1
            else:
                second = i+j
            index.append((second,first))
            j+=1
        deltas[i] = ( array[index[0][0]]-array[index[0][1]] + (2 * (array[index[1][0]]-array[index[1][1]])) ) / 10
    return deltas

def extract_features(audio,rate):

        try:
            """extract 20 dim mfcc features from an audio, performs CMS and combines 
            delta to make it 40 dim feature vector"""    

            # audio is audio signal from which to compute features -> should be n*1 array
            # rate is samplerate of the signal we are working with
            # 0.025 is the length of the analysis window in seconds (default is 25ms)
            # 0.01 is the step between successive windows in seconds (default is 10ms)
            # 20 is number of cepstrum to return (default is 13)
            # append energy is true if zeroth cepstral coefficient is replaced with log of total frame energy
            # mfcc() returns a numpy array of size (NUMFRAMES by numcep) containing features, each row holds 1 feature vector 
            # further possible parameters & their defaults can be found at python-speech-features.readthedocs.io/en/latest/
            mfcc_feat = mfcc.mfcc(audio,rate, 0.025, 0.01, 20, appendEnergy = True)

            # Scale all data onto one scale, eliminating sparsity & following same concept of Normalization & Standardization 
            mfcc_feat = preprocessing.scale(mfcc_feat)

            delta = calculate_delta(mfcc_feat)

            combined = np.hstack((mfcc_feat,delta))
            print("Features extracted")
            return combined
        except Exception as e:
            print(e)
            print("Extract features failed.")

它给我们的错误是：

后端终止（返回码：-9）

Answer 1

返回码：-9表示信号9，表示

SIGKILL。 SIGKILL信号发送到进程以使其立即终止（终止）。与SIGTERM和SIGINT相比，此信号不能被捕获或忽略，并且接收过程在接收到该信号后无法执行任何清除操作。

这可能就是为什么它没有被异常捕获的原因。

从音频中提取特征时如何修复（返回码：-9）

1 个答案: