In recent decades, the proliferation of parakeets has escalated into a significant and urgent problem for crops in the Rio de La Plata region. Their adaptability to different environments, high reproductive rates, and flexible diet have rapidly increased their population, posing a serious threat to agricultural productivity. This work is part of a crucial feasibility study for a parakeet detection system and subsequent deterrence based on real-time audio signals captured in the field. Recent studies have unveiled a fascinating aspect of parakeet behaviour-they possess a unique acoustic fingerprint. This allows them to encode their identity through different calls, enabling them to recognise each other and exchange information. This high level of acoustic complexity makes their sounds distinguishable from other field noises, a key feature that our proposed detection system aims to leverage. Our decision to employ and train a convolutional neural network (CNN) in this research was based on the recent breakthroughs of CNNs in classifying acoustic events. The input for this model is the spectrogram of audio samples, and it functions as a binary classifier, determining the presence or absence of parakeets. The results of our initial tests have been highly promising, bolstering our confidence in the potential of this system.