SADISS stands for socially aggregated, digitally integrated sound system.
It is a web-based application that bundles smartphones into monumental yet intricate sound systems or choirs/ensembles. It is developed at Institute for Composition, Conducting & Computer music of the Anton Bruckner University (Linz, Austria). The project's homepage you can find here: sadiss.net
Following a client-sever model, the SADISS server distributes information about sounds or text to be synthesized to the smartphones/clients currently registered, which in turn synthesize sound or speech accordingly. Clients connect to the server via the Internet, they do not have to be on the same Wifi, Subnet, etc.
Structure of SADISS' architecture:
User interaction with the SADISS client app is very basic. Users, after starting the app, are asked to scan a QR code to register in a SADISS performance. In case the particular performance requests user input regarding the selection of the language or manual sync setting ('expert mode') the user will have to set parameters accordingly. For most performances a simple scanning of the QR will take users into 'registered' mode.
<aside> <img src="/icons/note-sixteenth-beamed_purple.svg" alt="/icons/note-sixteenth-beamed_purple.svg" width="40px" /> To be able to hear SADISS tracks you will need to download and install the client apps to your smartphone(s). The SADISS client app is available from Google Play Store and Apple App Store.
</aside>
<aside> <img src="/icons/list_green.svg" alt="/icons/list_green.svg" width="40px" /> Groups Pages for SADISS PROJECT
Text to Speech Integration Instructions
Group Distributed Sound - Sound System
</aside>
Structure of SADISS' architecture:
There are two basic ways of working with SADISS, one we refer to the as Sound System, the other one as the Choir mode:
Example: For a concert/performance 100 people show up in the audience. Using the SADISS client app on their phones thes scan ONE single QR code displayed at the entrance to the venue to register their phones with the SADISS server which uses the 100 smartphones as individual, synchronized synthesizers creating a massively multichannel, immersive sea of sound.
Using TextToSpeech synthesis, SADISS can also synthesize on (potentially multi-lingual) stream of speech.
Structure of SADISS in sound system mode:
Example: 40 people show up to a performance. Using the SADISS app on their phones, they register via 4 different QR-codes. With 10 audience members having scanned each one of the 4 QR-codes, we have just created four sub-groups (or voices) of 10 members each.
Using headphones, the audience turned (co-)performers listen to pitches to sing/hum and (via text-to-speech synthesis) spoken instructions to follow. This potentially enables us to have the group of people sing 4-part harmonies and/or perform different actions while doing so.
In 'choir' mode for each SADISS voice an individual QR code is created:
‘choir’ mode establishes stable relationships between partial & TTS voice and individual client (determined by QR scanned):
While SADISS builds on one central approach, that is bundling of smartphones into orchestras by distributing different partials and text-to-speech items to be synthesized, quite a lot of different use cases ranging from music to theatre and public intervetion can be imagined.
It is important to understand that the SADISS server does not distribute audio-files but rather information about what frequencies and text snippets the individual client apps registered shall synthesize on the smartphones they are running on.
For Text-ToSpeech synthesis this means that we have NO control over the actual sound of the computer voice (male, female, accent, etc.). For audio synthesis this means we DO NOT KNOW the volume of each phone's contribution to the overall sound of the re-synthesized spectrum, nor can we know how its' speakers actually sound (some of them sound pretty rough). Also, difference in latency might be astonishing.
Hence we strongly advise novice users of SADISS NOT TO WORRY too much about phase correlation and other such minute details, but to embrace the chaotic wonders a group of people holding phones in physical space might discover.
https://camo.githubusercontent.com/fb2a6b440699b3e3b217912cbdcb60906c590c09be26ee79ce671168ed3a9e04/68747470733a2f2f7361646973732e6e65742f6170692f6c6f676f5f626c61636b2e706e67
SADISS on GitHub
Tracks contain the information necessary for audio and and Text-To-Speech synthesis. To create a new track you will have to upload a .txt file containing the partials to be synthesised by the clients as well as the .SRT-files for the Text-To-Speech synthesis. It is possible for a track to contain only partial- or TTS-information.
A SADISS performance is a collection of tracks, which can be can be started, stopped or looped from within it. The SADISS clients are linked to a specific performance via the QR-code(s) scanned at the venue.
Audio synthesis can use the following waveforms: sine, square, saw, triangle
How to create and edit partial files for audio synthesis: Sound system instructions
How to create SRT files: Text To Speech Instructions
In choir mode, you will have to upload partial- and SRT-files for each voice individually. For choir tracks containing more than one language, for each voice a seperate SRT file for each language will be required.
SADISS tracks are arranged in performances.
SADISS uses information encoded in QR-Codes to link a client's app to a specific performance. The QR codes to be scanned using the SADISS client apps are to be generated from within the 'performance' page of the performance interface.
SADISS QR-Codes contain information on:
More on the generation of QR-Codes within SADISS.
Audio: A SADISS audio track contains for every 10ms frequency/amplitude values for all currently active partials.
An example containing the information distributed to the registered clients in the first 30ms of a track:
0.000000 1 0 1567.367432 0.000391
0.010000 1 0 1566.510986 0.001300
0.020000 5 0 1565.874268 0.003313 4 4699.596191 0.000108 3 6270.289551 0.000134 2 7835.834961 0.000085 1 10969.394531 0.000137
In Sound System mode this information about partials to be synthesized is distributed dynamically to all registered clients.
In Choir mode, fixed relations between individual client and partial IDs are established. For this to work, partial files cannot be simple results from audio analysis, in which each partial is numbered continuously, but organise partial information accordingly.
Text-To-Speech (TTS) Synthesis: SADISS uses .SRT files to integrate speech synthesis. It is possible to upload different .SRTs containing text in different languages to enable multi-lingual performance. In choir mode utilising TTS, for each voice an individual SRT-file will need to be uploaded. For multi-lingual tracks multiple SRTs for each voice are needed.
User interaction with the SADISS client app is very basic. Users, after starting the app, are asked to scan a QR code to register in a SADISS performance. In case the particular performance requests user input regarding the selection of the language or manual sync setting ('expert mode') the user will have to set parameters accordingly. For most performances a simple scanning of the QR will take users into 'registered' mode.