why Python input overflowed error during adaptive

问题:

import pyaudio, math, struct
import numpy as np
import matplotlib.pyplot as plt

#inital config
chunk = 1024
FORMAT = pyaudio.paInt16
CHANNELS = 1
RATE = 44100
RECORD_SECOND = float(input('seconds:' ))

p = pyaudio.PyAudio()

guess = 0
integral = 0
kp = 0.5
ki = 200
kd = 9999
dt = RATE**(-1)
ddt = RATE
stream = p.open(format=FORMAT,
                channels=CHANNELS,
                rate=RATE,
                input=True,
                output=True,
                frames_per_buffer=chunk)

total_error = 0
previous_e = 0

#start processing
print("recording")

for i in range(0, int(RATE / chunk * RECORD_SECOND)):
    byted = (b'')
    #stream.read(chunk, exception_on_overflow = False)
    for element in np.fromstring(stream.read(chunk), 'Int16'):
        stream.read(chunk, exception_on_overflow = False)
        error = -(guess - element)
        integral += error*dt
        derivative = (error - previous_e) / ddt
        guess += int(round(kp*error + ki*integral + kd*derivative, 0))
        byted += struct.pack('<h', -guess)
        previous_e = error
    stream.write(byted, chunk)

#close
stream.stop_stream()
stream.close()
p.terminate()

input('press enter to exit...')    

If you run the above code, it will progress to the recording part.

"IOError: [Errno -9981] An Input Overflowed" error occurs.

I have tried to change various values ​​related to chunk and rate to see why this problem occurs, but the same thing always happens. I wonder if there is an answer to this question.