Reciepent of THE FUTURE OF THE PRESENT 2002 artist residency.
of Brecht Machine
Press Release for Brecht Machine at SPLIT Festival, Croatia
Proposal for Brecht Machine (EU Popstar)
CLICK HERE TO WATCH THE VIDEO
Machine by G.H.Hovagimyan- two-way live webcast/performance
Sept. 26th , online.
dictation and translation systems are imperfect, expect quite a bit of
poetic mayhem and unexpected translation goofs!!
Proposal for Brecht Machine (EU Popstar)
The idea is simple. Use the Internet to stream audio and video. The audio portion will be spoken in one language and translated into another language at the receiving end. At the receiving end the voice speaking will be synthetic. This will be accomplished in this way. A dictation program will translate the spoken dialog into text. The text will be passed to a translation program and translated. The translated text will be spoken by a synthetic voice. This will be done by stringing together a series of programs to accomplish each task and automating the whole process using a scripting language such as AppleScript for Macintosh and Visual Basic Script for PC.
The first step of the process uses dictation software. This involves a person training the software to recognize their voice and the way they use words. The diction software is limited artificial intelligence. The person as to communicate and interact with this limited A.I.. At first the software program misunderstands many works but through patient training it can achieve a 99% recognition rate. The mistakes and misinterpretations are what interests me. This is a fertile area for art that would be similar to the Surrealists idea of Automatic Writing. I can discover a new way of speaking and writing that is more suitable to conversing with an artificial intelligence.
The second step in this process is the translation software. This is usually word by work machine translation. This too can create misinterpretation and delightful linguistic surprises. It also has a way of limiting the type and style of language so that the software may translate correctly.
These parameters create an interesting arena for the development fo an artificial media language, a manner of speaking and writing that is suitable for use over the internet and across the many different languages. Just as radio and television have created a certain type of language, speaking and writing most effective for those mediums it is entirely appropriate to create new language forms for future internet audiences that are multicultural and multilingual. Another interesting aspect of this process is the combination of two types of media entities. One is the familiar televised person the other is the computer synthetic voice. This is the basis for a new type of hybrid performer.
My proposed teleplay/performance over the internet contains other creative disjunctions. It is assumed that the translation process will not synchronize exactly with the audio video stream. This creates a texture to the presentation in which the spoken text is separated from the moving image. By separating out these two elements a new way of viewing the action is created for the end receiver. Creating a performance art work within this new structure is quite interesting. It allows for a critical distance of the viewer within the actual play.
All images: CICV 2000 Interview, Peter Sinclair and GH Hovagimyan
First I create a script that strings together the software. Here is the structure. A person speaks into a video camera and microphone. The audio signal is split. Once part goes to the streaming software which encodes the video and audio for transmission over the internet. This is simple teleconferencing or video streaming such as Cu See Me, NetMeeting, Quicktime 5 or RealMedia. The split audio signal goes to an audio mixer that goes to a computer. The computer meets dictation software (speech to text) to create a text document. The text document is the sent over the internet to a receiving computer which has the translation software and the synthetic voice running locally. The local accessor hears what the actor is saying in their own native tongue. Obviously this is a simplistic example. But it highlights the various aspects of what needs to be created.
There are two different models one may use in creating instant translation. One is a broadcast model. A person accesses a channel that has the teleplay translated into their language. This is all done from one location so that the dictation -> translation -> voice synthesis stays on the server side. The other way is to distribute the computing tasks to local hard drives so that one part of the process, the dictation and transmission over the internet is done by the server and the other part, the translation and voice syntheses is done by the client on their local hard drive. I favor distributing the modules locally. The task then becomes creating the translation -> voice syntheses modules so that people may download and install these modules on their local hard drives.
three modules to start, English, French and German. I feel that enlisting
three separate locations to present this work is the proper approach.
The three proposed partners are; Ecole Normal Superieur de Telecommunications,
Paris or the CICV (International Center for the Creation of Video) in
France, The University of Dortmund or the ZKM in Germany, and Parsons
School of Design (Franklin Furnace). I am in talks with all of the locations
at present and will have a better idea of the exact partners a little
further along in the process.