Acacia

This work explores how technologies that augment
our understanding of bird populations in order to allow these populations
to speak to us about their habitat. In particular, in this collaboration
between the MIT Media Laboratory and the Maine Audubon Society, we use
cellular technology augment the process by which a group of volunteers
collect information for an OWL census in Maine.

The methodology was developed in a pilot census of Connecticut's owl
population, conducted in the summer of 2006. In this study, we
demonstrated that the audio quality of cell phones was sufficient for the
discovery and interaction with owls. In this project, the cell phone is
used to make the owl call and to record the response. The success of the
pilot suggests that this small, portable technology can replace the
conventional high-quality audio survey broadcasting and recording
equipment.

In our work in Maine we will deploy more cell nodes for calling owls and
recording their response. We anticipate that each deployed node will
result in to be several hours of recording per night per phone. In order
to not overwhelm, individual volunteers who participate in the annotation
of these audio streams, we are building a collaborative community that
will assist in the task. A scheduling, annotation and database website
will provide the backbone necessary for the volunteers to conduct and
evaluate their experiments and, at the same time, will allow the public to
participate in the processes of interpretation and annotation.

With this project, we hope to gain insight into the social networking
processes of collaborative interpretation and annotation of a shared
database; knowledge representation for the bird-census domain; and the
design issues involved in creating and maintaining a website for community
scientific collaboration.

The cellular survey may also provide data which suggest insights into
questions about the hearing range of owls, duplication of vocalizing
individual responses in adjacent experiment sites, the response rate of
owls due to current weather or human presence, and comparison between
trigger-based and naturally occurring responses in surveys. In addition,
specific signal processing and communication technologies will be
field-tested.

Machine harmonic transcription of Jazz music investigates algorithmic methods for mimicking human understanding of Jazz harmonies. It combines the short-term set-estimation of musical chords, the modeling and analysis of chord sequences patterns, and a higher-level understanding of Jazz. Jazz music is particularly challenging in that it purposefully accommodates dissonance and ambiguity in enriching the improvisational environment.