Data analysis and machine-learning algorithms are used to make predictions about our needs and desires. Various sensing devices such as voice-controlled assistants are embedded within our daily lives, using microphones, cameras and data caches to capture our behaviour, building unique profiles based on our habits, purchases and biometric data (body-specific data, like a fingerprint, heart rate, or voice).
This new data landscape is a key subject for Deep Voice. Empty Orchestra. We investigate the link between entertainment, being in the ‘limelight’ and the vulnerable position of using your own body and voice in our current data-centric world. The result is an installation that acts as a karaoke booth. The title ‘Deep Voice. Empty Orchestra.’ refers to the Japanese roots of the terms, kara 空 ‘empty’ and ōkesutora オーケストラ ‘orchestra’, combined with the term ‘deep’ as a reference to deep-learning; a form of machine-learning that can learn from and adapt to vast amounts of data. A good example is DeepFace, an advanced image recognition tool used by Facebook that can independently learn to differentiate and accurately identify thousands of people in pictures.
Our interest in the practice began while Guillemette was living in Taiwan and heard about a conspiracy theory which asserts that karaoke is being used by government-owned biometric data companies—such as China-sponsored companies iFlyTek, D-ear, and SinoVoice—to process hundreds of millions of uniquely identifiable voices. While EU General Data Protection Regulation (GDPR) prohibits this type of biometric data processing for the purpose of uniquely identifying a person unless there is a lawful basis under Article 6, in the UK, the HMRC (tax authority) was found to be unlawfully storing more than 5 million taxpayers’ biometric voiceprints without consent.
For this research we partnered with the Netherlands-based (Rhenen) software company Good Vibrations Company B.V., which has created a software that is able to analyse real-time emotions detected through voice, as well as the general mood of the person singing.
Through an engaging karaoke installation, Legrand Jäger invites visitors to sing along and see their voice analysis mapped in real-time, allowing vital insight into the data that a single voice can produce, and the often unseen back-end of data generation and analysis.