One way to slice a wav file in memory and supply to scipy wavreader

slice_wav_in_memory.py

from pydub import AudioSegment
from io import BytesIO
from scipy.io import wavfile

# get file
audio_file = AudioSegment.from_wav("some.wav")

# create output, supply to AudioSegment
audio_sample = BytesIO()
audio_file[2000:5000].export(audio_sample, format='wav')

# read into scipy.wavreader
audio_info = wavfile.read(BytesIO(audio_sample.getvalue()))