What is the Speaker Detector Module?
The Speaker Detector module is a simple module that listens to a microphone input (mono or stereo) and detects when auditory activity has happened. It is mostly intended for situations in which two users interact side-by-side with a system, in order to detect which one of them is speaking. It outputs a message with the following Active Speaker state: None, Left, Right or Both. In order to provide a smoother output, detection follows an Attack/Release method to detect changes, and thus minimize false-positives. In single-user situations, it can still be used with a mono microphone and a mono audio-interface – in that case the module will always detect right and left-user activity together, meaning that the message None will mean silence, and the message Both will mean that the single user is speaking (with a mono audio-interface, the mono signal is copied into both stereo channels, thus the system will detect activity in both).
How it was used in emote?
This module was used in Scenario 2 – Enercities to let the system know when and which player was speaking, in order to both adapt gazing behaviour, and also simple turn-taking behaviour. It was used with two lavalier-microphones connected to a stereo USB audio-interface (Zoom H6). A microphone was placed on each player.
What, if any, other software/hardware is needed?
- Mono or stereo microphone / audio-interface
Download Speaker Detector: http://gaips.inesc-id.pt/emote/downloads/speaker-detector/