convolution-dsp 0.3.0

1-dimensional convolution library intended for use in DSP applications.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29

#!/usr/bin/env python3
import numpy as np
from scipy.signal import oaconvolve
import time

kernel = np.float32(np.random.rand(4001))

re = np.float32(np.random.rand(2**15))
im = np.float32(np.random.rand(2**15))
signal = np.array(list(map(lambda v: v[0] + v[1] * 1j, zip(re, im))))
signal = np.complex64(signal)

for _ in range(10):
    t1 = time.time()
    output = oaconvolve(kernel, signal)
    t2 = time.time()
    print(f"Convolution took {t2 - t1} seconds")
output = np.complex64(output)

with open("kernel2.bin", "wb") as f:
    kernel.tofile(f)

with open("signal2.bin", "wb") as f:
    signal.tofile(f)

with open("output2.bin", "wb") as f:
    output.tofile(f)

print(signal[10])