This section describes the typical layout of a talk room at a DebConf. It includes the video and audio components. An online conference has all required functions remotely operated.
The video hardware we use is described in the video hardware section. There are two cameras in every room and an Opsis board for capturing the presenter’s presentation. These inputs are mixed together using Voctomix. The Opsis output is streamed over the wired network by the capture PC to Voctomix. The output from Voctomix is streamed out to the stream servers and recorded locally for upload to the review system.
The audio setup described here uses the equipment listed in the hardware section. It uses:
two headset mics for presenters
two handheld mics for questions
two omni-directional mics for ambient noise
one stereo DI for the presenter’s laptop audio
Ambient noise is the clapping and general buzz that happens in the background during a talk. This is useful for the stream, as it makes viewers feel as though they are in the room. These two mics are not sent to the room audio system. The presenter’s laptop audio is captured to ensure that any audio or video played during the talk can also be played on the stream and PA.
There are two audio mixes which need to be somewhat separate - one for the PA and one for the stream. Most desks can support this using an aux bus or sub-groups. Using aux buses requires 2 post-fader buses (left and right). This is so the channel fader affects both the PA and the stream levels. The main-mix, or one of the sub-groups, is run to the camera for the stream and recording. This way, the camera handles synchronizing the audio and video. The aux bus or other sub-group pair is run to the room’s speakers.
Combined, this looks like this:
Voctomix is controlled via a web interface by the Director. It mixes between the various sources that are streamed out to the audience. The Talkmeister and presenters communicate in a Jitsi call. This is streamed through Voctomix for the Q&A section of pre-recorded talks and the full talk for BoFs and live events.
Recordings are uploaded to sReview for transcoding
Recordings are then played back through vogol
Jitsi is used for Q&A and BoFs streaming rooms
Jibri is used to record Jitsi streams. We have two instances to have a fallback
EtherpadLite is used for people to ask questions during Q&A and for taking notes during BoFs
Voctomix is the live video mixing software we use
Vogol is a Voctomix front end
We use RTMP to stream at low latency
We also use HLS to stream, as it has wider compatibility even though it comes with higher latency