RoSE: A Research-oriented Social Environment

"RoSE is a research-oriented social environment for tracking and integrating relations between authors and documents in a combined 'social-document graph.' It allows users to learn about an author or idea from the evolving relationships between people-and-documents, people-and-people, and documents-and-documents. . . . RoSE is currently a demonstration project in early development by the UC Transliteracies Project, which focuses on the digital reading in today's socially-networked digital environments." (UC Transliteracies website) The RoSE development team includes graduate student programmers, artists, and literary researchers at UCSB. The project is directed by Professor Alan Liu and co-directed by Professor Rita Raley and Rama Hoetzlein.

The interactive visualization user interface for RoSE was developed by Salman Bakht in Adobe Flex. Salman is currently developing the "storyboard" environment, which allows users to create their own visualizations.

Intermission (MATLAB software)

Software for Intermission was developed in MATLAB by Salman Bakht. The software algorithmically generates the audio tracks for the installation from a set of audio loops. Although this software was not used in the original exhibition, it allows for rapid generation of the audio for the installation in new environments.

www.lovelyweather.com (software)

Software for www.lovelyweather.com was developed in Adobe Flex by Salman Bakht. The software algorithmically generates a continuous animation, compositing hundreds of photos, and a multitrack audio stream in real time.

PAPAGEI

PAPAGEI (Parametric Algorithmic Patchbay-Accompanist Generating Events by Improvisation) is an automatic accompanist developed by Salman Bakht and Clarence Barlow. The software is written in C++ using the JUCE Library.

Version 1 of the software analyzes microphone input from a flute performer and generates an accompanying melody for (sampled) flute. Using a combination of pitch detection (based on autocorrelation) and loudness detection, instantaneous input performance parameters (pitch/register, density, loudness) are mapped to several output parameters (density, metricity, tonality), while specified melodic sequences trigger changes in others (increase and decrease of tempo, increase and decrease of melodic range). These output parameters are translated into a randomly generated melody using probability tables based on melodic, harmonic, and rhythmic theories of Clarence Barlow.

A modified version of PAPAGEI was used in the piece PIANO 04h by Salman Bakht. This version of the software allows for multiple simultaneous input pitches received as MIDI notes and multiple output melodies to be generated.

CONFETTI

CONFETTI is a random sample-playback/granulation software synthesizer written by Salman Bakht in C++ using the JUCE Library. Modified versions of this software were used in a number of artistic works including Nodes and Passages, Mushrooms, and _object.fragment.

Samples are loaded into an array of sample banks corresponding to a range of MIDI notes (pitches). The samples in each bank can then be subdivided to any length. Optionally, an envelope can be applied to the samples. When a MIDI note is triggered, a sequence of randomly chosen subsamples from the appropriate sample bank are played until the note is released. By default, the subsample is chosen by a uniform probability function. The probability distribution may also be set to a bell-shaped distribution (a sum of uniformly distributed values), where the user can control the variance and mean of the distribution via sliders.

Implementation of a Time-Domain Auditory Model

This project was developed by Salman Bakht and Agata Sikorski under the advisement of Mike Davies at King's College London in the summer of 2001.

Summary: A time-domain auditory model based upon that described by David J. M. Robinson and Malcolm J. Hawksford in their paper "Time-domain auditory model for the assessment of high-quality coded audio" is implemented. Possible uses for this model are predicting the perceived quality of coded audio or for testing imperceptibility of audio watermarking. This model utilizes a non-linear filter bank to simulate spectral analysis as performed by the inner ear and complex filtering of the envelope to simulate the temporal response of the ear. Tests were performed to demonstrate the spectral and temporal masking effects caused by the ear, showing the model works generally as expected.