Subtlenoise produces low-impact but information-rich soundscapes in realtime. The intended usecase is to monitor distributed computing operations and supplement the traditional monitoring services such as dashboards.
The implementation is described below but is flexible in terms of the various components which may be used. Subtlenoise is more of an architecture than a complete software solution. At the heart of the system is a python script used to sonify incoming messages and output messages suitable for audio-rendering.
We want monitoring to be less intrusive. The well-known pattern of time-series histogram is visually rich but also visually overwhelming when digesting many such plots. This project attempts to use sonification to provide an Auditory display.
An audio stream is less intrusive if done correctly (and horribly annoying if done poorly). Ultimately we want a situation where the ambient sounds can be useful and totally unobstrusive (Picard knows)
The human brain is adept at extracting information from subtle environmental noise (Paul knows)
The following components are being used to collect, transport, translate, and emit messages in the Subtlenoise pipeline:
- UDP (collection)
- mqtt (messaging)
- python (orchestration and translation of messages)
- SonicPI (audio rendering)
Opensource sound rendering tool is SonicPi which uses Supercollider under the hood. Most synth engines will accept Open Sound Control (OSC) messages as output by the python orchestration script. To do this we need the python-osc and mqtt packages. Install via pip:
$ pip install paho-mqtt
$ pip install python-osc
- use dj-mqtt.py to subscribe to the mqtt message stream
- consume messages using the dj-mqtt.py script
- start Sonic Pi and ensure the OSC server is running
- experiment with Sonic Pi ruby code (example is render.rb)
-
The abscence of signal is informative
-
A common pattern is to use audio cues when certain events occur, clumsily called Earcons. For example, the train whistle when the LIGOS experiment loses beam lock (Earcon example).
-
Needs to be unobtrusive otherwise it becomes very annoying very quickly
-
Monitorrama talk from Joe Ruscio with some toptips for visualisation http://vimeo.com/62630749
-
Nice description of how to process incoming data thelisteningmachine
-
pertinent comment "...more than just white noise. It is a richly layered audio experience, with enough variation to prevent monotony but not so variable as to become distracting (which ambient sound should never be)” - http://www.youtube.com/watch?v=icFT76pHgoM
- NOT entertainment BBC sounds of space
- NOT musical LHCchamber music
- NOT a curiosity LHC sound
The above are good examples of outreach and inspirational news items, but are not the sonification we’re looking for...