E-Locus - Institutional Repository of the University of Crete - Mixing and reproduction of user generated audio recordings

Home Mixing and reproduction of user generated audio recordings

Results - Details

[Add to Basket]

Identifier

000406877

Title

Mixing and reproduction of user generated audio recordings

Alternative Title

Ψηφιακές τεχνικές μίξης και αναπαραγωγής ηχογραφήσεων προερχόμενες από πληθοπωρισμό

Author

Βισκαδούρος, Μενέλαος Μ.

Thesis advisor

Μουχτάρης, Αθανάσιος

Reviewer

Τσακαλίδης, Παναγιώτης
Αργυρός, Αντώνιος

Abstract

We live in the era of smart multimedia devices, drones and smartphones, devices capable of capturing every moment of our lives and of the public events that we attend. Recordings from these devices, produced by many users participating in the same public event, become available through the social media and the large number of websites which provide video and audio content. This way, they become easily accessible to the general public and to the scientific community. In the current work, we investigate different means by which this rich material can be processed and combined, so as to produce an audio stream which provides an enhanced listening experience of the captured acoustic event. Considering a collection of User Generated audio Recordings (UGR) which are organized and synchronized along a common time axis, we explore various simple monophonic and stereophonic mixing techniques, exploiting all the recordings which are available at each time instance. Moreover, we introduce Maximum Component Elimination (MCE), a novel multichannel signal processing technique which is implemented in the time - frequency domain with the goal to suppress acoustic noise and interference in the near field of each recording device. A listening test was conducted with the scope to study the preference of the listener with respect to material resulting from the mixture of UGRs from different devices, compared to material originating from a single recording device. The results indicate that some of the mixing techniques gather significantly higher preference in comparison to when UGRs are reproduced individually. Additional listening tests were conducted with the scope to evaluate the ability of MCE in suppressing noise and interference which is unique to each recording location. The results of the test indicate that unwanted noise, such as speech or wind, is much less perceived or not perceive d at all when using MCE, as opposed to when the UGRs are simply superimposed.

Language

English

Subject

Growdsourcing

Maximun Component Elimination (MCE)

Mixing techniques

Quality features

UGR

Ηχογραφήσεις

Περιεχόμενο παραγόμενο από το χρήστη

Τεχνικά χαρακτηριστικά

Issue date