Voice recognition technology is such an interesting tool for performance. I imagine using it to make a dating app inspired website where you find your “type” through voicing out keywords.
I started building the project by altering the in-class example. I couldn’t find an API revenant to “hot dudes” so I decided to hand curate some images. Due to time restrain, I used p5.speech to prototype after several failed attempts to apply grammar specification to webkitSpeechRecognition.
The downside of p5.speech’s recognition is that it tends to pick up the entire sentence instead of just the keyword and hence it is a bit unnatural to perform with it. In the ideal setting, I will be describing what I am looking for in sentences one after the other and the image will switch accordingly. I am really interested in this project and will continue experimenting and making it work with Chrome’s webkitSpeechRecognition. Rather than one image at a time, I would want the project to start with multiple images and have user eliminate the ones without the input keywords.
Moving forward, I played with Surrealist Siri which is unlike the conventional Siri but “an intelligent assistant to help you free words from the constraints of rational and discursive order”.
As shown in the screencast of a short conversation between Surrealist Siri and myself, Surrealist Siri’s responds were nothing related to what I was talking about. In the beginning, I thought that maybe Surrealist Siri will give me a more relatable responds the more I talked to it. However, its remained poetic but random. I find the concept fascinating as it stripped away the purpose of an assistive voice technology device. Unlike a conventional Siri where you have a correlated conversation, Surrealist Siri is strangely distant.