Academic literature on the topic 'Active speaker detection'

Speaker diarization systems aim to find ‘who spoke when?’ in multi-speaker recordings. The dataset usually consists of meetings, TV/talk shows, telephone and multi-party interaction recordings. In this paper, we propose a novel multimodal speaker diarization technique, which finds the active speaker through audio-visual synchronization model for diarization. A pre-trained audio-visual synchronization model is used to find the synchronization between a visible person and the respective audio. For that purpose, short video segments comprised of face-only regions are acquired using a face detection technique and are then fed to the pre-trained model. This model is a two streamed network which matches audio frames with their respective visual input segments. On the basis of high confidence video segments inferred by the model, the respective audio frames are used to train Gaussian mixture model (GMM)-based clusters. This method helps in generating speaker specific clusters with high probability. We tested our approach on a popular subset of AMI meeting corpus consisting of 5.4 h of recordings for audio and 5.8 h of different set of multimodal recordings. A significant improvement is noticed with the proposed method in term of DER when compared to conventional and fully supervised audio based speaker diarization. The results of the proposed technique are very close to the complex state-of-the art multimodal diarization which shows significance of such simple yet effective technique.

Spoken language is fundamentally different from the written language in that it contains frequent disfluencies or parts of an utterance that are corrected by the speaker. Disfluency detection (removing these disfluencies) is desirable to clean the input for use in downstream NLP tasks. Most existing approaches to disfluency detection heavily rely on human-annotated data, which is scarce and expensive to obtain in practice. To tackle the training data bottleneck, in this work, we investigate methods for combining self-supervised learning and active learning for disfluency detection. First, we construct large-scale pseudo training data by randomly adding or deleting words from unlabeled data and propose two self-supervised pre-training tasks: (i) a tagging task to detect the added noisy words and (ii) sentence classification to distinguish original sentences from grammatically incorrect sentences. We then combine these two tasks to jointly pre-train a neural network. The pre-trained neural network is then fine-tuned using human-annotated disfluency detection training data. The self-supervised learning method can capture task-special knowledge for disfluency detection and achieve better performance when fine-tuning on a small annotated dataset compared to other supervised methods. However, limited in that the pseudo training data are generated based on simple heuristics and cannot fully cover all the disfluency patterns, there is still a performance gap compared to the supervised models trained on the full training dataset. We further explore how to bridge the performance gap by integrating active learning during the fine-tuning process. Active learning strives to reduce annotation costs by choosing the most critical examples to label and can address the weakness of self-supervised learning with a small annotated dataset. We show that by combining self-supervised learning with active learning, our model is able to match state-of-the-art performance with just about 10% of the original training data on both the commonly used English Switchboard test set and a set of in-house annotated Chinese data.

Past research suggests that the ability to recognise the emotional intent of a speaker decreases as a function of age. Yet, few studies have looked at the underlying cause for this effect in a systematic way. This paper builds on the view that emotional prosody perception is a multi-stage process and explores which step of the recognition processing line is impaired in healthy ageing using time-sensitive event-related brain potentials (ERPs). Results suggest that early processes linked to salience detection as reflected in the P200 component and initial build-up of emotional representation as linked to a subsequent negative ERP component are largely unaffected in healthy ageing. The two groups show, however, emotional prosody recognition differences: older participants recognise emotional intentions of speakers less well than younger participants do. These findings were followed up by two neuro-stimulation studies specifically targeting the inferior frontal cortex to test if recognition improves during active stimulation relative to sham. Overall, results suggests that neither tDCS nor high-frequency tRNS stimulation at 2mA for 30 minutes facilitates emotional prosody recognition rates in healthy older adults.

In this paper, we present a novel approach referred to as the audio-based virtual landmark-based HoloSLAM. This innovative method leverages a single sound source and microphone arrays to estimate the voice-printed speaker’s direction. The system allows an autonomous robot equipped with a single microphone array to navigate within indoor environments, interact with specific sound sources, and simultaneously determine its own location while mapping the environment. The proposed method does not require multiple audio sources in the environment nor sensor fusion to extract pertinent information and make accurate sound source estimations. Furthermore, the approach incorporates Robotic Mixed Reality using Microsoft HoloLens to superimpose landmarks, effectively mitigating the audio landmark-related issues of conventional audio-based landmark SLAM, particularly in situations where audio landmarks cannot be discerned, are limited in number, or are completely missing. The paper also evaluates an active speaker detection method, demonstrating its ability to achieve high accuracy in scenarios where audio data are the sole input. Real-time experiments validate the effectiveness of this method, emphasizing its precision and comprehensive mapping capabilities. The results of these experiments showcase the accuracy and efficiency of the proposed system, surpassing the constraints associated with traditional audio-based SLAM techniques, ultimately leading to a more detailed and precise mapping of the robot’s surroundings.