Segmenting audio into homogeneous sections such as music and speech helps us understand the content of audio. It is useful as a pre-processing step to index, store, and modify audio recordings, radio broadcasts and TV programmes. Deep learning models for segmentation are generally trained on copyrighted material, which cannot be shared. Annotating these datasets is time-consuming and expensive and therefore, it significantly slows down research progress. In this study, we present a novel procedure that artificially synthesises data that resembles radio signals. We replicate the workflow of a radio DJ in mixing audio and investigate parameters like fade curves and audio ducking. We trained a Convolutional Recurrent Neural Network (CRNN) on this synthesised data and outperformed state-of-the-art algorithms for music-speech detection. This paper demonstrates the data synthesis procedure as a highly effective technique to generate large datasets to train deep neural networks for audio segmentation.

Haptic feedback is used in cars to reduce visual inattention. While tactile feedback like vibration can be influenced by the car's movement, thermal and cutaneous push feedback should be independent of such interference. This paper presents two driving simulator studies investigating novel tactile feedback on the steering wheel for navigation. First, devices on one side of the steering wheel were warmed, indicating the turning direction, while those on the other side were cooled. This thermal feedback was compared to audio. The thermal navigation lead to 94.2% correct recognitions of warnings 200m before the turn and to 91.7% correct turns. Speech had perfect recognition for both. In the second experiment, only the destination side was indicated thermally, and this design was compared to cutaneous push feedback. The simplified thermal feedback design did not increase recognition, but cutaneous push feedback had high recognition rates (100% for 200 m warnings, 98% for turns).

The use of ultrasound haptic feedback for mid-air gestures in cars has been proposed to provide a sense of control over the user's intended actions and to add touch to a touchless interaction. However, the impact of ultrasound feedback to the gesturing hand regarding lane deviation, eyes-off-the-road time (EORT) and perceived mental demand has not yet been measured. This paper investigates the impact of uni- and multimodal presentation of ultrasound feedback on the primary driving task and the secondary gesturing task in a simulated driving environment. The multimodal combinations of ultrasound included visual, auditory, and peripheral lights. We found that ultrasound feedback presented unimodally and bimodally resulted in significantly less EORT compared to visual feedback. Our results suggest that multimodal ultrasound feedback for mid-air interaction decreases EORT whilst not compromising driving performance nor mental demand and thus can increase safety while driving.

This paper presents an investigation into the effects of different feedback modalities on mid-air gesture interaction for infotainment systems in cars. Car crashes and near-crash events are most commonly caused by driver distraction. Mid-air interaction is a way of reducing driver distraction by reducing visual demand from infotainment. Despite a range of available modalities, feedback in mid-air gesture systems is generally provided through visual displays. We conducted a simulated driving study to investigate how different types of multimodal feedback can support in-air gestures. The effects of different feedback modalities on eye gaze behaviour, and the driving and gesturing tasks are considered. We found that feedback modality influenced gesturing behaviour. However, drivers corrected falsely executed gestures more often in non-visual conditions. Our findings show that non-visual feedback can reduce visual distraction significantly

This demonstration showcases novel multimodal feedback designs for in-car mid-air gesture interaction. It explores the potential of multimodal feedback types for mid-air gestures in cars and how these can reduce eyes-off-the-road time thus make driving safer. We will show four different bimodal feedback combinations to provide effective information about interaction with systems in a car. These feedback techniques are visual-auditory, auditory-ambient (peripheral vision), ambient-tactile, and tactile-auditory. Users can interact with the system after a short introduction, creating an exciting opportunity to deploy these displays in cars in the future.

Infotainment Systems can increase mental workload and divert visual attention away from looking ahead on the roads. When these systems give information to the driver, providing it through the tactile channel on the steering wheel might improve driving behaviour and safety. This paper describes an investigation into the perceivability of haptic feedback patterns using an actuated surface on a steering wheel. Six solenoids were embedded along the rim creating three bumps under each palm. A simulated driving study was conducted to test for recognition accuracy of the haptic patterns (81.3%). There was no significant increase in lane deviation or steering angle during haptic pattern presentation. These results suggest that drivers can reliably distinguish between cutaneous patterns presented on the steering wheel. Our findings can assist in delivering non-critical messages to the driver (e.g. driving performance, incoming text messages, etc.) without decreasing driving performance or increasing perceived mental workload.