2.0.0a5 - read_audio reads old voices

[onurgule]

Hello,

I am using read_audio function to do something then I respond with write_audio.

   while pcall.state == call.CallState.ANSWERED:
        if silence_counter >= max_silence_count:
            break
        data = pcall.read_audio(length=800, blocking=True)
        if data and ((data.count(b"\x80") + data.count(b"\x7f")) / len(data)) <= 0.94:
            w.writeframes(data)
        
        time.sleep(1 / (8000 / 800))

After then, I respond with write_audio due to command:

    f = wave.open(file, 'rb')

    frames = f.getnframes()
    data = f.readframes(frames)
    f.close()
    print("FRAMES:",frames)

    pcall.write_audio(data)  # This writes the audio data to the transmit buffer
    print("Data are in buffer and playing")

    stop = time.time() + (frames / 8000)  # frames/8000 is the length of the audio in seconds

    while time.time() <= stop and pcall.state == call.CallState.ANSWERED:
        time.sleep(0.1)
    print("Message finished")

But, when I interrupt and speak while write_audio running, it detects after writing finished and reading starts.
For example:

Script: Hello, how can I help you?
Me: Turn on the lights.
Script: Alright, I am turning... (I INTERRUPT IT and say "No, Close the lights") ... the lights. Anything else?
Me: Thank you.
Script: (It reads audio like "No, Close the lights thank you"): Okay, I am closing the lights.

I used 1.6.7 and 2.0.0a5 and both versions didn't work.
I want to read audio only when I want to read, not previous voices.

Also, I tried blocking=False but it didn't work and also has many delay because of writing.

Thanks.