Audio Labels Research Articles

Auxiliary audio–textual modalities for better action recognition on vision-specific annotated videos

Most current audio–visual datasets are class-relevant, where audio and visual modalities are annotated. Thus, current audio–visual recognition methods apply cross-modality attention or modality fusion. However, leveraging the audio modality effectively in vision-specific videos for human activity recognition is of particular challenge. We address this challenge by proposing a novel audio–visual recognition framework that effectively leverages audio modality in any vision-specific annotated dataset. The proposed framework employs language models (e.g., GPT-3, CPT-text, BERT) for building a semantic audio–video label dictionary (SAVLD) that serves as a bridge between audio and video datasets by mapping each video label to its most K-relevant audio labels. Then, SAVLD along with a pre-trained audio multi-label model are used to estimate the audio–visual modality relevance. Accordingly, we propose a novel learnable irrelevant modality dropout (IMD) to completely drop the irrelevant audio modality and fuse only the relevant modalities. Finally, for the efficiency of the proposed multimodal framework, we present an efficient two-stream video Transformer to process the visual modalities (i.e., RGB frames and optical flow). The final predictions are re-ranked with GPT-3 recommendations of the human activity classes. GPT-3 provides high-level recommendations using the labels of the detected visual objects and the audio predictions of the input video. Our framework demonstrated a remarkable performance on the vision-specific annotated datasets Kinetics400 and UCF-101 by outperforming most relevant human activity recognition methods.

Medical Image Description Based on Multimodal Auxiliary Signals and Transformer

Medical image description can be applied to clinical medical diagnosis, but the field still faces serious challenges. There is a serious problem of visual and textual data bias in medical datasets, which are the imbalanced distribution of health and disease data. This can greatly affect the learning performance of data-driven neural networks and finally lead to errors in the generated medical image descriptions. To address this problem, we propose a new medical image description network architecture named multimodal data-assisted knowledge fusion network (MDAKF), which introduces multimodal auxiliary signals to guide the Transformer network to generate more accurate medical reports. In detail, audio auxiliary signals provide clear abnormal visual regions to alleviate the visual data bias problem. However, the audio modality signals with similar pronunciation lack recognizability, which may lead to incorrect mapping of audio labels to medical image regions. Therefore, we further fuse the audio with text features as the auxiliary signal to improve the overall performance of the model. Through the experiments on two medical image description datasets, IU-X-ray and COV-CTR, it is found that the proposed model is superior to the previous models in terms of language generation evaluation indicators.

Accessible Point-and-Tap Interaction for Acquiring Detailed Information about Tactile Graphics and 3D Models.

We have devised a novel "Point-and-Tap" interface that enables people who are blind or visually impaired (BVI) to easily acquire multiple levels of information about tactile graphics and 3D models. The interface uses an iPhone's depth and color cameras to track the user's hands while they interact with a model. When the user points to a feature of interest on the model with their index finger, the system reads aloud basic information about that feature. For additional information, the user lifts their index finger and taps the feature again. This process can be repeated multiple times to access additional levels of information. For instance, tapping once on a region in a tactile map could trigger the name of the region, with subsequent taps eliciting the population, area, climate, etc. No audio labels are triggered unless the user makes a pointing gesture, which allows the user to explore the model freely with one or both hands. Multiple taps can be used to skip through information levels quickly, with each tap interrupting the current utterance. This allows users to reach the desired level of information more quickly than listening to all levels in sequence. Experiments with six BVI participants demonstrate that the approach is practical, easy to learn and effective.

Accessible interactive 3D models for blind and low-vision people

Blind and low-vision (BLV) people experience difficulty accessing graphical information, particularly regarding travel and education. Tactile diagrams and 3D printed models can improve access to graphical information for BLV people; however, these formats only allow limited detailed and contextual information. Interactive 3D printed models (I3Ms) exist, but many rely on passive audio labels that don't particularly empower BLV people in independent knowledge building and interpretation. This project investigates the creation of I3Ms that offer more engaging experiences with a focus on facilitating independent exploration and knowledge discovery. Specifically, this project explores how BLV people want to interact with I3Ms, interactive functionalities and behaviours that I3Ms should support, such as conversational interfaces and model agency, and to understand the relationship between I3Ms and conventional accessible graphics.

A Human-in-the-Loop System for Sound Event Detection and Annotation

Labeling of audio events is essential for many tasks. However, finding sound events and labeling them within a long audio file is tedious and time-consuming. In cases where there is very little labeled data (e.g., a single labeled example), it is often not feasible to train an automatic labeler because many techniques (e.g., deep learning) require a large number of human-labeled training examples. Also, fully automated labeling may not show sufficient agreement with human labeling for many uses. To solve this issue, we present a human-in-the-loop sound labeling system that helps a user quickly label target sound events in a long audio. It lets a user reduce the time required to label a long audio file (e.g., 20 hours) containing target sounds that are sparsely distributed throughout the recording (10% or less of the audio contains the target) when there are too few labeled examples (e.g., one) to train a state-of-the-art machine audio labeling system. To evaluate the effectiveness of our tool, we performed a human-subject study. The results show that it helped participants label target sound events twice as fast as labeling them manually. In addition to measuring the overall performance of the proposed system, we also measure interaction overhead and machine accuracy, which are two key factors that determine the overall performance. The analysis shows that an ideal interface that does not have interaction overhead at all could speed labeling by as much as a factor of four.

A Low-Cost Audio Prescription Labeling System Using RFID for Thai Visually-Impaired People

This research aims to develop a low-cost audio prescription labeling (APL) system for visually-impaired people by using the RFID system. The developed APL system includes the APL machine and APL software. The APL machine is for visually-impaired people while APL software allows caregivers to record all important information into the APL machine. The main objective of the development of the APL machine is to reduce costs and size by designing all of the electronic devices to fit into one print circuit board. Also, it is designed so that it is easy to use and can become an electronic aid for daily living. The developed APL software is based on Java and MySQL, both of which can operate on various operating platforms and are easy to develop as commercial software. The developed APL system was first evaluated by 5 experts. The APL system was also evaluated by 50 actual visually-impaired people (30 elders and 20 blind individuals) and 20 caregivers, pharmacists and nurses. After using the APL system, evaluations were carried out, and it can be concluded from the evaluation results that this proposed APL system can be effectively used for helping visually-impaired people in terms of self-medication.

Real-time speech-driven animation of expressive talking faces

In this paper, we present a real-time facial animation system in which speech drives mouth movements and facial expressions synchronously. Considering five basic emotions, a hierarchical structure with an upper layer of emotion classification is established. Based on the recognized emotion label, the under-layer classification at sub-phonemic level has been modelled on the relationship between acoustic features of frames and audio labels in phonemes. Using certain constraint, the predicted emotion labels of speech are adjusted to gain the facial expression labels which are combined with sub-phonemic labels. The combinations are mapped into facial action units (FAUs), and audio-visual synchronized animation with mouth movements and facial expressions is generated by morphing between FAUs. The experimental results demonstrate that the two-layer structure succeeds in both emotion and sub-phonemic classifications, and the synthesized facial sequences reach a comparative convincing quality.

SPEECH/MUSIC DISCRIMINATION BASED ON WARPING TRANSFORMATION AND FUZZY LOGIC FOR INTELLIGENT AUDIO CODING

Automatic discrimination of speech and music is an important tool in many multimedia applications. This article presents an evolutionary, fuzzy, rules-based speech/music discrimination approach for intelligent audio coding, which exploits only one simple feature, called Warped LPC-based Spectral Centroid (WLPC-SC). Comparison between WLPC-SC and the classical features proposed in the literature for audio classification is performed, aiming to assess the good discriminatory power of the proposed feature. The length of the vector for describing the proposed psychoacoustic-based feature is reduced to a few statistical values (mean, variance, and skewness), which are then transformed to a new feature space, applying linear discriminant analysis (LDA), with the aim of increasing the classification accuracy percentage. The classification task is performed applying a support vector machine (SVM) to the features in the transformed space. The final decision is made by a fuzzy expert system, which improves the accuracy rate provided by the SVM, taking into account the audio labels assigned by this classifier to past audio frames. The accuracy rate improvement due to the fuzzy expert system is also reported. Experimental results reveal that our speech/music discriminator is robust and fast, making it suitable for intelligent audio coding.

Audio-visual sports highlights extraction using Coupled Hidden Markov Models

We present our studies on the application of Coupled Hidden Markov Models(CHMMs) to sports highlights extraction from broadcast video using both audio and video information. First, we generate audio labels using audio classification via Gaussian mixture models, and video labels using quantization of the average motion vector magnitudes. Then, we model sports highlights using discrete-observations CHMMs on audio and video labels classified from a large training set of broadcast sports highlights. Our experimental results on unseen golf and soccer content show that CHMMs outperform Hidden Markov Models(HMMs) trained on audio-only or video-only observations. Next, we study how the coupling between the two single-modality HMMs offers improvement on modelling capability by making refinements on the states of the models. We also show that the number of states optimized in this fashion also gives better classification results than other number of states. We conclude that CHMMs provide a promising tool for information fusion techniques in the sports domain for audio-visual event detection and analysis.