Voice is becoming a stronger and stronger player in our digital presence; voice navigation, voice assistants, Alexa, Google Home, Siri, etc... The visual interface is disappearing. Big corporations are busy creating better and better synthetic voices that resemble a real human voice. For this purpose, large amounts of voice data is collected. This data is processed using sophisticated algorithms to synthesise new voices.
However, a voice (our own voice) is something very personal:
“Voice simultaneously produces articulate sound and produces myself, as a self-producing being” Steven Connor, 2000. So a strange new phenomenon is emerging; a synthetic voice without a self. While artificial speech still lacks most of the subtleties that characterise our voices when speaking, insecurity pride rage, happiness etc. "Ehms" and "Aha's" are already convincingly replicated by Google Duplex.
Repeat After Me is a permanent interactive installation in one of the former telephone booths of Haus der Kulturen der Welt in Berlin. After picking up the telephone, the visitor is asked to repeat some words, sentences and sounds. These sounds are mixed with previous sound recordings of previous visitors and played back as a sound piece inside the booth. Also a twitterbot tweets short audio clips from the piece every now and then.
On a technical note, we’ve outfitted the telephone with an Arduino to detect pickups. A small Processing sketch on the computer attached to the Arduino is monitoring these events and forwarding them via websockets to a Chrome browser running the interactive part of the installation. The audio processing is being handled by Node.js using the audio and video libraries FFmpeg and SoX.