Thanks to Scott Panfil from for his help with the audio engineering!

Since 1976, Thesis Audio Service has provided quality, affordable repairs, maintenance, and restorations on all types of equipment for musicians, sound companies,recording studios,and music stores.
We are an authorized factory service center for over 35 major brands of amplifiers, keyboards, guitars, and sound gear. We offer guitar repair, vintage amplifier restorations, guitar amplifier repair, keyboard repairs, vintage synthesizer repair and restorations, audio equipment repairs, vintage audio repairs and restorations, custom work and modifications.

 Legacy speakers anchor the Raven Audio system

Designed for Thesis Audio Turntables, but can also be used with other makes of turntable, CD players, amplifiers and pre-phono units to increase mass and isolate equipment from vibration in the listening environment. Provided with three conical leveling feet.
Other sizes available by special order.

Material: Non-resonating stone
Dimensions: 50x35 cm
Weight: approx. 10 Kg

Thesis Audio Service amplifier repairs

Roseman joined Morning Edition to break down his process for this year's mashup; hear the full story at the audio link.

In this thesis we propose the architecture of a modular system that allows closely related people, such as relatives and friends, to maintain awareness of a distant site, such as home, through auditory cues and eavesdropping. While people are at work or traveling, they often want to have some degree of awarereness about activities related to family members' health and well-being. The solution that we propose is a light infrastructure system that listens into the home and in response to a change in activity (e.g. cooking, taking a shower, watching TV) sends a few seconds of audio to the caregivers providing them with background awareness of events occurring at the remote site. In order to protect privacy at the monitored site the system is designed to classify sounds and then send audio icons, pre-recorded pieces of sound representing activities (e.g. a recording of a baby crying), instead of the actual sound. However, when the classifier's confidence level is low and speech is detected, the system sends the actual sound but in a garbled way. In the absence of speech, the system simply sends the actual sound. The system takes classifying decisions based on acoustic evidence and possible location of the source. Several clients can connect to the system and simultaneously receive alerts.