Audio monitoring information technology plays an important role in the application of monitoring systems, and it is an indispensable and important link. Whether intelligent audio monitoring management can be successfully realized, the key is to successfully detect abnormal sounds from a variety of external environment background sounds. The core technology of abnormal sound detection is a pattern classification task. The dimension of features is fixed in the traditional abnormal sound detection model. Such an ordinary solution will lead to a long time-consuming detection process and increase the boundary error. Traditional speech detection is not good enough for sound discrimination in a noisy environment, so this paper proposes an abnormal speech detection technology based on moving edge computing. Aiming at the noisy environment of the music classroom, the determination of objective function should be further optimized. Through the related technology, a certain sound can be quickly identified and analyzed in the music classroom to promote the development of the music wisdom classroom, and music wisdom classrooms can be used as a computer-aided system to help music teachers better grasp the learning situation of students, put forward relevant guidance strategies, improve students’ learning enthusiasm, and enhance teachers’ teaching efficiency so as to promote the progress of music teaching.

Urban acoustic environments and performing-arts spaces strongly influence speech communication quality, acoustic comfort, and public wellbeing, particularly in noise-exposed shared environments such as transport hubs, campuses, healthcare spaces, public service facilities, music-education settings, and rehearsal or performance-related spaces. To address speech–noise separation in low signal-to-noise ratio and acoustically complex scenarios, this study proposes a lightweight two-stage deep learning framework termed LCGRU–Wave-SkipConvNet. In the preprocessing stage, a Lightweight Convolutional Gated Recurrent Unit (LCGRU) model is employed to achieve preliminary separation of target speech and background noise by capturing both spatial and temporal acoustic features. In the post-processing stage, a Wave-SkipConvNet model is introduced to further suppress residual noise and enhance speech quality. Experimental results demonstrate that the proposed framework achieves superior performance under different signal-to-noise ratios, sound-source angles, and target angle errors. For example, in the preprocessing stage, the LCGRU model achieved a perceptual evaluation of speech quality (PESQ) score of 2.64 at source angles between 0° and 30°, outperforming the convolutional neural network-long short-term memory (CNN-LSTM) model by 1.17. In the post-processing stage, the Wave-SkipConvNet model achieved higher short-time objective intelligibility (STOI) and segmental signal-to-noise ratio (segSNR) values than the comparison models under different SNR conditions. The proposed framework provides an effective and deployment-oriented AI solution for improving speech accessibility and acoustic comfort in urban acoustic environments and performing-arts spaces. Beyond speech enhancement, it offers practical potential for supporting healthier, more inclusive, and more equitable acoustic environments in noise-sensitive public and educational spaces. It should be noted that this study focuses on the objective acoustic environment and signal-level speech enhancement, rather than subjective soundscape perception, musical perception, or human perceptual evaluation.