The aim of this paper is to propose hybrid content radio, a new framework for radio content, enhancing the traditional broadcast radio experience and augmenting it with context related audio content.
Differently from most of the commercial recommendation-based internet streaming services (Spotify, Pandora), here we consider systematically adding audio content to an existing, linear audio structure.
The purpose of the hybrid content radio framework is to enhance the broadcaster’s programme schedule with context-aware and personalised audio content from the internet.
The context can be the listener’s profile, emotional state and activity, her geographical position, the weather, and all factors contributing to characterize the state of the listener.
The final purpose of the enhancement is to improve the service user’s’ listening experience, decreasing their propensity to channel-surf and giving them more targeted content, such as local news, entertainment, music and also relevant advertisements.
In this way, the hybrid content radio approach enables both a functional enhancement to radio and network resource optimization, allowing the use of both the broadcast channel and the internet.
This paper gives an overview of the recent experimental services proposed by a group of European Broadcasters exploring the potentialities of a hybrid approach for audio in radio.
In hybrid content radio (HCR), traditional linear broadcast radio is the foundation upon which a new, enriched service is built, using enriching audio content from the broadcaster’s archives or from trusted third party providers.
The paper presents experimental services and outlines key technical requirements for the creation of an HCR radio framework.
Different from most of the internet streaming services, here we consider adding audio content to an existing, linear audio structure: the broadcaster’s programme schedule.
Specifically, HCR allows enhancement of the broadcaster’s linear schedule with context-aware and personalised audio content.
The context can be the listener’s emotional state, her geographical position, her group, the weather and all factors contributing to it .
The final purpose of the enhancement is to improve the service user’s listening experience, giving her more targeted contents, such as news, entertainment, music and also relevant advertisements.
The proposed technique achieves content personalisation at a minimal bandwidth cost, as the broadcast channel is used if possible, differently from existing internet music playlists. In this way, HCR allows an optimized bandwidth usage.
Figure 1 illustrates the concept: broadcast linear audio content is enriched by personalised content from the internet.
The proposed framework can be applied to both audio and video content. However, audio is well suited as a background medium, and can be enjoyed while people are doing something else.
It’s common to see people listening to radio while walking, biking or driving or engaged in different activities. In this sense, context has a more complex impact on radio than on television.
There are already several mobile music streaming services creating highly personalised playlists, exploiting different content recommendation techniques: Pandora, based on content features extracted by experts, the Music Genome Project , others mainly based on collaborative filtering or hybrid techniques like Spotify .
Music streaming services generally use recommender systems exploiting collaborative filtering, content-based or social-based techniques [4-6] and exclusively use the internet channel to reach listeners with wholly customized playlists.
Different from those services, hybrid content radio addresses all the scenarios where the linear audio content is partly and flexibly replaced by personalised audio content, and the broadcaster maintains overall control.
Several IST European Projects have considered enhancing broadcast video content, recombining broadcast objects at the client and downloading them from the broadband network.
Specifically, the iMedia and Savant IST Projects [7-9] used targeted, personalised advertisements or recommended content lists for television services; however context- awareness was not considered.
In past years, music and video recommenders generally suggested personalised playlists and dynamic content only focusing on user or item- similarity, the customer profile and navigation history.
In contrast, recent research and commercial services have started addressing contextual information, leveraging the context such as the user’s position, mood or activity .
Most recently, a number of studies analysed the proposition of context-aware audio content.