Using the interaural time difference and cross-correlation to localise short-term complex noises

The mammalian binaural cue of interaural time difference (ITD) and cross-correlation have long been used to determine the point of origin of a sound source. The ITD can be defined as the different points in time at which a sound from a single location arrives at each individual ear [1]. From this time difference, the brain can calculate the angle of the sound source in relation to the head [2]. Cross-correlation compares the similarity of each channel of a binaural waveform producing the time lag or offset required for both channels to be in phase with one another. This offset corresponds to the maximum value produced by the cross-correlation function and can be used to determine the ITD and thus the azimuthal angle θ of the original sound source. However, in indoor environments, cross-correlation has been known to have problems with both sound reflections and reverberations. Additionally, cross-correlation has difficulties with localising short-term complex noises when they occur during a longer duration waveform, i.e. in the presence of background noise. The crosscorrelation algorithm processes the entire waveform and the short-term complex noise can be ignored. This paper presents a technique using thresholding which enables higher-localisation abilities for short-term complex sounds in the midst of background noise. To determine the success of this thresholding technique, twenty-five sounds were recorded in a dynamic and echoic environment. The twenty-five sounds consist of hand-claps, finger-clicks and speech. The proposed technique was compared to the regular cross-correlation function for the same waveforms, and an average of the azimuthal angles determined for each individual sample. The sound localisation ability for all twenty-five sound samples is as follows: average of the sampled angles using cross-correlation: 44%; cross-correlation technique with thresholding: 84%. From these results, it is clear that this proposed technique is very successful for the localisation of short-term complex sounds in the midst of background noise and in a dynamic and echoic indoor environment.

Dithen: A Computation-as-a-Service Cloud Platform For Large-Scale Multimedia Processing

We present Dithen, a novel computation-as-a-service (CaaS) cloud platform specifically tailored to the parallel ex-ecution of large-scale multimedia tasks. Dithen handles the upload/download of both multimedia data and executable items, the assignment of compute units to multimedia workloads, and the reactive control of the available compute units to minimize the cloud infrastructure cost under deadline-abiding execution. Dithen combines three key properties: (i) the reactive assignment of individual multimedia tasks to available computing units according to availability and predetermined time-to-completion constraints; (ii) optimal resource estimation based on Kalman-filter estimates; (iii) the use of additive increase multiplicative decrease (AIMD) algorithms (famous for being the resource management in the transport control protocol) for the control of the number of units servicing workloads. The deployment of Dithen over Amazon EC2 spot instances is shown to be capable of processing more than 80,000 video transcoding, face detection and image processing tasks (equivalent to the processing of more than 116 GB of compressed data) for less than $1 in billing cost from EC2. Moreover, the proposed AIMD-based control mechanism, in conjunction with the Kalman estimates, is shown to provide for more than 27% reduction in EC2 spot instance cost against methods based on reactive resource estimation. Finally, Dithen is shown to offer a 38% to 500% reduction of the billing cost against the current state-of-the-art in CaaS platforms on Amazon EC2 (Amazon Lambda and Amazon Autoscale). A baseline version of Dithen is currently available at dithen.com.