House mice (Mus musculus), like other rodents, communicate using sonic and ultrasonic vocalizations (USVs), but their functions are still poorly understood. One of the main challenges for studying any acoustic communication is processing and analyzing audio files. Our aims here are to provide a critical and comprehensive review of the new bioacoustic tools available for processing and analyzing recordings of mouse vocalizations. We consider each method as used in a serial data processing pipeline and how to minimize errors at each step to prevent error propagation (or error cascades). First, we review methods for processing audio files of recordings of mice. We compare conventional approaches for visualizing vocalizations (time-frequency representations) with an alternative method adapted to the mouse auditory system. We compare machine learning (ML) and signal processing methods for automating USV detection and emphasize the need for better methods for denoising audio files and reliable frequency contour (ridge) tracking and feature extraction. Second, we review methods for analyzing detected USVs, focusing on classification and sequencing approaches. Classifying USVs is a challenging task because, while some calls are discrete, others show graded variation within and between call classes. We compare supervised classifications and unsupervised labeling, and we emphasize the importance of reliable manual (researcher-based) methods as a gold standard for automated ML approaches. We review classifications of mouse vocalizations in the literature, and we propose a new hierarchical framework for the classification of USVs. We examine methods for sequencing USVs and consider their relative advantages. Finally, we address the unresolved technological challenges for these methods to study rodent vocalizations and propose potential solutions for the future.

Bioacoustic recordings are often degraded by ambient noise, which complicates the analysis of weak or noise-overlapped vocalizations. Convolutional neural networks, particularly U-Net architectures, have shown strong denoising performance in speech and music processing. However, their application for denoising bioacoustic signals is limited by the scarcity of clean training data. To address this, we propose a training set synthesis approach and develop a supervised denoising model that predicts a complex ratio mask in the time-frequency domain. The model leverages ridges, or frequency contours, that represent the fundamental frequency together with one or more harmonic partial components of vocalizations. These ridges are used both for the synthesis of training set and for the design of a loss function that assigns higher weights to the ridge regions (ridge-guided loss function). This weighting step helps the network better preserve vocalization details during denoising. As a case study, we evaluate our approach using ultrasonic vocalization (USV) recordings of house mice, which are widely studied in behavioral biology and neuroscience. In field recordings, the proposed method enhances tracking of fundamental and harmonic partial ridges compared to previous signal-processing approaches. In addition, a classifier trained on denoised data improves USV classification on out-of-sample, noisy recordings from wild and domesticated mice compared to classifiers trained on noisy recordings. The proposed method also substantially improves the scale-invariant signal-to-distortion ratio on synthetic testing data across a wide range of input signal-to-noise ratios. While focused on USVs, the proposed approach is broadly applicable to other bioacoustic signals with trackable ridges, and thus enables ridge-based training set synthesis and denoising.