MIT Media Lab colorbar
MIT Media Lab

     Jerk-O-Meter: Speech-Feature Analysis Provides Feedback on Your Phone Interactions

     VIEW: Research Profile
     DEVELOPED BY: Anmol Madan, Research Associate, MIT Media Lab
     ADVISOR: Alex (Sandy) Pentland, Toshiba Professor of Media Arts and Sciences, MIT Media Lab
     RESEARCH GROUP: Human Dynamics
     ASSOCIATED RESEARCH: GroupMedia Project

The Jerk-O-Meter The Jerk-O-Meter The Jerk-O-Meter
Click on photos for high-resolution versions (300dpi) suitable for print.
All photos this page: Webb Chappell.

Profile By Anmol Madan, Research Associate
Have you ever had the experience where you call someone up and he or she doesn't seem to be paying attention to you?

The Jerk-O-Meter (or JerkoMeter) is a real-time speech feature analysis application that runs on your VOIP phone or cell phone that remedies precisely that experience. It uses speech features that measure activity and stress (and soon empathy) from your tone of voice and speaking style, to predict if you are 'being a jerk' on the phone. The phone displays appropriate messages, and can also be set up to inform the person on the other end of the line that you're extremely busy. The messages range from "Stop being a jerk!" to "Wow, you're a smooth talker," based on your performance. The application is currently designed to analyze only the user's conversation, and not the person at the other end of the line. The Jerk-O-Meter is the work of Anmol Madan, a PhD candidate at the MIT Media Laboratory, and Dr. Alex (Sandy) Pentland, a pioneer in wearable and socially aware computing.

Anmol Madan with the Jerk-O-Meter
Anmol Madan and the Jerk-O-Meter

The mathematical models for the Jerk-O-Meter were derived from several research studies at the Media Lab. These studies evaluated how a person's speaking style could reflect his or her interest in a conversation, in going out on a date, or perhaps even in buying a particular product. Our results show that a person's speaking style and 'tone of voiceĠ can predict objective outcomes (e.g. interest in a conversation, or in going out on a date) with 75-85% accuracy. For more academic information, please refer to the links at the end of this page.

The Jerk-O-Meter is just one possible implementation of the underlying technology, which has much broader applications in areas such as customized user experiences, consumer research, advertising, marketing, movies and television audiences, call centers, and various consumer and corporate applications. An important implication is that computers and cell phones may now understand people better, just as other people do. Instead of users adapting their habits to work with computers or cell phones, these devices could support us in the ways that we naturally communicate.

The current version of the Jerk-O-Meter is a research prototype, and runs in Linux on the Zaurus VOIP phone. Ron Caneel, alum of the research group, wrote part of the code to extract the activity and stress levels in real time. From a technical perspective, the Jerk-O-Meter could easily be converted into a downloadable application for a cell phone.

Jerk-O-Meter is an academic research project and is not commercially available at this time.

Academic Papers about Jerk-O-Meter Research
"Voices of Attraction": Research paper describing the speech features in more detail. (PDF, 6 pages)
GroupMedia project site: Describes various studies in more detail.
"Social Dynamics: Signals and Behavior": A paper which provides more background on the speech features of the research. (PDF, 5 pages)
Human Dynamics group publications site: Describes related work in the group.

Future Downloads:
Interested in downloading Jerk-O-Meter software for your cell phone, mobile device, or computer?
If so, please provide your device information, and we will contact you when this software is available.
Your information will only be used for software notification.


Your email address:

Info needed: Cell phone handset (manufacturer, model),
PDA, VOIP software, and/or other platform:

Press Contact:
Alexandra Kahn, Press Liaison, MIT Media Lab
akahn **at** or 617 253-0365

Project Contact:
Anmol Madan
anmol **at**

MIT Media Lab