Question

我目前正在处理一个项目，其中我需要使用音频文件的光谱。

我目前正在使用另一个程序来生成音频文件的功率谱，并使用matplotlib绘制它，如下：

for elements in file_names_data_start_end_index:
    ## Entries = Rows of the output data
    ## Length  = Columns of the output data
    entries = elements[2] - elements[1]+1
    name = elements[0]
    length = data_splitted[elements[1]].count(" ")-2
    print "Filename: " + name + " Entries: " + str(entries) +  " Length: "  + str(length)
    output = ""
    for i in range(elements[1],elements[2]+1):
        output += data_splitted[i]
        #print data_splitted[i]
    #print output[:-1]
    output = np.fromstring(output[:-1],count=-1,sep=' ')
    #output_log = np.log(output)
    output_reshaped = output.reshape(entries,length)
    #print output_reshaped.shape
    ##Do the actual plot!
    # 1 second audio creates 98 rows, as there is a overlap on 10 ms will
    # extra data be created, normally would it have been 16000/512 = 31 rows
    # if the frames where disjoint, but not in this case. 160 samples is the extra overlap.
    # (16000-512)/160+1 = 97.8 ~ 98
    Y =  np.array(range(0,length)) * (16000.0/512.0)
    X =  np.array(range(0,entries))*(1/98.0)
    X,Y = np.meshgrid(X, Y)
    #print type(X)
    #print type(Y)
    #print type(output_reshaped)
    plt.pcolormesh(X,Y,output_reshaped.T,  cmap=cm.jet)
    #surf = ax.plot_surface(X,Y,output_reshaped)
    plt.xlabel('Time(s)')
    plt.ylabel('Frequency(Hz)')
    plt.title('Power spectrum of ' + name)
    plt.colorbar()
    plt.savefig(spectogram+"test_spect_plots/"+name+"_plot.png")
    plt.close()
    print "Saved"

问题在于，即使音频文件的时间更长或更短，情节总是具有相同的大小......

enter image description here

时间轴的分辨率不固定，有没有办法保持它固定，这样matplotlib不会尝试将图像拟合到前缀大小，而只是扩展x轴并使绘图有点更长？

创建具有不同x轴长度的谱图

0 个答案: