Fraunhofer IDMT’s automatic Speech and Music Detection licensed by German public Broadcasters

Press Release /

Until now, reporting and accounting of musical content used in TV shows, spots, intro sequences and as background music in talk-shows and TV interviews was only based on rough estimates. Given that these estimates have a critical impact on the remuneration of the right holders, Fraunhofer IDMT has developed a new technology that allows automatic and efficient detection of speech and music content in the daily or weekly TV broadcasts. The public German TV broadcasters ARD and ZDF now rely on Fraunhofer IDMT’s music and speech detection technology to analyze their channels in the German Broadcasting Archive.

Amongst others, IDMT´s technologies for automatic A/V analysis help to detect errors and quality issues in video content.

Fraunhofer IDMT´s automatic music and speech detection is an efficient and reliable solution to detect all TV spots with musical content. The method is based on a universal algorithm for classifying audio signals, which has been trained using machine learning techniques. It can be used to divide files and broadcast streams reliably and statistically accurately into segments ‚with music‘ and ‚without music‘.

The results of such analysis can be used by the licensing department to determine segments ‚with music‘ in the TV program that do not have a precise track report. A request for the missing track report can be then made to the corresponding editorial department.

Steffen Holly, head of business unit Media Management & Delivery at Fraunhofer IDMT, proudly explains: “This is an important advance and the right step towards fairer remuneration for involved musicians and rights holders”.

At this year´s IBC in Amsterdam being on September 11 until 15, Fraunhofer IDMT will present this new software tool for automatic music and speech detection in hall 8 booth 8.B80.

Fraunhofer IDMT will also present its technologies for the automatic A/V analysis that enable efficient management of digital media content for broadcasters and archives.

With the automatic Quality Control (QC), errors and quality issues in A/V material can be detected at any stage of the production process.

The Video Segment Matching software automatically detects video content or parts of videos used in the production process. The automatic detection is possible even if the content has been manipulated or transcoded. This saves time, allows quick and easy reuse of the material (such as in online media libraries), and facilitates the accurate accounting of copyright protected content.

Recording devices and editing steps leave characteristic traces in audio material. IDMT shows at the IBC its Audio Forensics Tools to detect editing and tampering processes.

The automatic Annotation Tools detect individual shots, movements in video files, recording devices used or transcodings and provide this as additional information for the analyzed file.

Fraunhofer IDMT will also present its Actor Recognition technology that allows to identify individuals such as actors, celebrities, moderators, or athletes in video content.

Come to the Fraunhofer IDMT booth and experience how efficient management of digital media content can be achieved with our automated analysis technologies.