谁能指导我如何从 wav 文件中计算以 hz 为单位的频率？ 波形文件为 50 秒

Question

I am using numpy library to calculate freq = np.fft.rfftfreq(len_data, 1.0 / rate), If I am not wrong then this frequency is without unit. How can i convert it into hertz . I am using the following code:

import numpy as np
import scipy.io.wavfile
from scipy import signal

def read_wav_file(file_name):
    sample_rate, Data_audio  = scipy.io.wavfile.read(file_name)
    return sample_rate, Data_audio 

def getFFT(Data_audio, sample_rate):
    len_data = len(Data_audio)
    Data_audio = Data_audio * np.hamming(len_data)
    fft = np.fft.rfft(Data_audio)
    fft = np.abs(fft)
    ret_len_FFT = len(fft)
    freq = np.fft.rfftfreq(len_data, 1.0 / sample_rate)
    return ( freq[:int(len(freq))], fft[:int(ret_len_FFT)], ret_len_FFT )

sample_rate_rec, Data_audio_rec = read_wav_file('2020rec.wav')

frequency_rec, fft_rec, ret_lenFFT_rec = getFFT(Data_audio_rec, sample_rate_rec)

print("frequency_rec: " + str(frequency_rec) )

**Output**
frequency_rec: [0.00000000e+00 8.33481508e-03 1.66696302e-02 ... 2.39999833e+04
 2.39999917e+04 2.40000000e+04]


# To convert frequencies into float format
np.set_printoptions(formatter={'float':'{:f}'.format})
print("frequency_rec: " + str(frequency_rec) )

**Output**
frequency_rec: [0.000000 0.008307 0.016614 ... 23999.983386 23999.991693 24000.000000]

Answer 1

The numpy FFT package has a built-in function to calculate the frequency vector to go along with your FFT output. Note that scipy outputs the sample rate wile numpy wants the sample spacing so you must invert it first.

import numpy as np

def getFrequencies(Data_audio, sample_rate_rec):
    return np.fft.fftfreq(n=len(Data_audio), d=1/sample_rate)

Since the output of scipy.io.wavfile.read claims to be in samples/sec, the output of np.fft.fftfreq will be in cycles/sec aka Hertz.