Improve Recognition Robustness Research Articles

Comparing detection accuracy of mountain chickadee (Poecile gambeli) song by two deep-learning algorithms

The use of autonomous recording units (ARUs) has become an increasingly popular and powerful method of data collection for biological monitoring in recent years. However, the large-scale recordings collected using these devices are often nearly impossible for human analysts to parse through, as they require copious amounts of time and resources. Automated recognition techniques have allowed for quick and efficient analysis of these recordings, and machine learning (ML) approaches, such as deep learning, have greatly improved recognition robustness and accuracy. We evaluated the performance of two deep-learning algorithms: 1. our own custom convolutional neural network (CNN) detector (specialist approach) and 2. BirdNET, a publicly available detector capable of identifying over 6,000 bird species (generalist approach). We used audio recordings of mountain chickadees (Poecile gambeli) collected from ARUs and directional microphones in the field as our test stimulus set, with our custom detector trained to identify mountain chickadee songs. Using confidence thresholds of 0.6 for both detectors, we found that our custom CNN detector yielded higher detection compared to BirdNET. Given both ML approaches are significantly faster than a human detector and the custom CNN detector is highly accurate, we hope that our findings encourage bioacoustics practitioners to develop custom solutions for targeted species identification, especially given the availability of open-source toolboxes such as Koogu.

Eidetic recognition of cattle using keypoint alignment

Ability to identify individual cows quickly and readily in the barn would enable real time monitoring of their behavior, health, eating habits and more, all of which could save time, money, or effort. This work focuses on creating an eidetic recognition or re-identification (ReID) algorithm that learns to recognize individual cows with just a single training example per cow and with near zero time to learn to identify a new cow, both features which the existing cattle ReID systems lack. Our algorithm is designed to improve recognition robustness to deformations in cow bodies that occur when they are walking, turning, or are seen slightly off-angle. Individual cows are first detected and localized using popular keypoint and mask detection techniques, then aligned to a fixed template, pixelated, binarized to reduce lighting effects, and serialized to obtain bit-vectors. Bit-vectors from cows at inference time are matched to those from training time using Hamming distance. To improve results, we add modules to verify the validity of detected keypoints, interpolate missing keypoints, and combine predictions from multiple frames using a majority vote. The video level accuracy is over 60% for a set of nearly 150 Holstein cows.

Using Object Reconstruction as a Dynamic Attention Window to Improve Recognition Robustness

Using Object Reconstruction as a Dynamic Attention Window to Improve Recognition Robustness

An Uncertainty Propagation Approach to Robust ASR Using the ETSI Advanced Front-End

In this paper, we show how uncertainty propagation, combined with observation uncertainty techniques, can be applied to a realistic implementation of robust distributed speech recognition (DSR) to improve recognition robustness furthermore, with little increase in computational complexity. Uncertainty propagation, or error propagation, techniques employ a probabilistic description of speech to reflect the information lost during speech enhancement or source separation in the time or frequency domain. This uncertain description is then propagated through the feature extraction process to the domain of features used in speech recognition. In this domain, the statistical information can be combined with the statistical parameters of the recognition model by employing observation uncertainty techniques. We show that the combination of a piecewise uncertainty propagation scheme with front-end uncertainty decoding or modified imputation improves the baseline of the advanced front-end (AFE), the state of the art algorithm of the European Telecommunications Standards Institute (ETSI), on the AURORA5 database. We compare this method with other observation uncertainty techniques and show how the use of uncertainty propagation reduces the word error rates without the need for any kind of adaptation to noise using stereo data or iterative parameter estimation.

Acoustic features and a distance measure that reduce the impact of training–test mismatch in ASR

For improved recognition robustness in mismatched training–test conditions, the application of key ideas from missing feature theory and robust statistical pattern recognition in the framework of an otherwise conventional automatic speech recognition (ASR) system were investigated. To this end, both the type of features used to represent the speech signals and the algorithm used to compute the distance measure between an observed feature vector and a previously trained parametric model were studied. Two different types of feature representations were used: a type in which spectrally local distortions are smeared over the entire feature vector and a type in which distortions are only smeared over part of the feature vector. In addition, two different distance measures were investigated, viz., a conventional distance measure and a robust local distance function in the form of acoustic backing-off. The effects on recognition performance were studied for artificially created, band-limited noise and NOISEX noise added to the speech signals. The results for artificial band-limited noise indicate that a partially smearing feature transform is to be preferred over a fully smearing transform. In addition, for artificial, band-limited noise, a robust local distance function is to be preferred over the conventional distance measure as long as the distorted feature values are outliers with respect to the feature distribution observed during training. The experiments with NOISEX noise show that the combination of feature type and distance measure that is optimal for artificial, band-limited noise is also capable of improving recognition robustness for NOISEX noise, provided that it is band-limited.

Dynamic auditory representations and statistical speech recognition

The most common spectral estimation algorithm used for automatic speech recognition incorporates rough approximations of basic aspects of auditory modeling: frequency selectivity and magnitude compression. Attempting to improve the robustness and overall performance of ASR, researchers have proposed more sophisticated auditory models as the spectral estimation front end, with generally modest success at best. One common concern throughout these efforts is that the representation derived from an auditory model may not be a good match for typical statistical recognition algorithms. Recently, a dynamic auditory model that emphasizes changing local spectral peaks and has improved recognition robustness compared to other common front ends has been derived and implemented. The present work uses specific case examples to show how the perceptual representation leads to a softening of the resulting statistical models. The works also proposes a simple mechanism to adapt dynamic spectral features into a form more suitable for segmentally static statistical characterization. The mechanism is based on approximating the temporal derivative of the frequency position of local spectra peaks. The impact of our auditory model with this processing mechanism on robust recognition performance is discussed. [Work supported by NIH Grant No. 1 R29 DC 02033-01A1, and by NSF.]