Talk-o-Meter shows, how much everybody is talking.

The left picture was taken during the learning phase of the "red" speaker. the center and the right picture were taken during actual conversations. Before starting the conversation, each partner says a few sentences during a brief learning phase. The app analyzes the frequency patterns of the voices and decides on how to tell the speakers apart. This calibration can be repeated, for example, if one of the conversation partners changes. The quality of the voice detection depends on how different the voices of the conversation partners are. Loud background noise reduces detection quality.

Voice detection is best when people speak naturally during the learning phase. The three center buttons set the interval in minutes, after which a new bar appears. Voice detection happens in realtime. If you chose a three minute interval, every three minutes a new bar appears, showing how much each speaker talked during the past interval. So you are getting feedback while the conversation is happening.

The app does not make a permanent recording/soundfile of the conversation, so people don't have to worry about privacy.

...shows, how much everybody is talking

Manual here	There are friends, colleagues or even partners who do not notice it, when they dominate others in a conversation. They just talk too much. A little iPhone app helps. The iPhone is placed between two people having a conversation and learns to separate the two voices. At intervals of 1, 2 or 5 minutes you see different lengths of red and blue bars that show what percentage of time each speaker was talking. Nobody has to be unpleasantly exhorted - from time to time everyone will have a cursory glance at the Talk-o-Meter and adapt if he is talking too much. Gentle biofeedback works. Notes: Talk-o-meter only works with 2 people who do not have similar voices and talk in a quiet environment. A version for more than 2 people is to follow shortly.
This iPhone App is available at the AppStore. The left picture was taken during the learning phase of the "red" speaker. the center and the right picture were taken during actual conversations. Before starting the conversation, each partner says a few sentences during a brief learning phase. The app analyzes the frequency patterns of the voices and decides on how to tell the speakers apart. This calibration can be repeated, for example, if one of the conversation partners changes. The quality of the voice detection depends on how different the voices of the conversation partners are. Loud background noise reduces detection quality. Voice detection is best when people speak naturally during the learning phase. The three center buttons set the interval in minutes, after which a new bar appears. Voice detection happens in realtime. If you chose a three minute interval, every three minutes a new bar appears, showing how much each speaker talked during the past interval. So you are getting feedback while the conversation is happening. The app does not make a permanent recording/soundfile of the conversation, so people don't have to worry about privacy. Talk-o-Meter is a product of Unperfekthaus, Inh. Reinhard Wiesemann, Friedrich-Ebert-Str. 18, 45127 Essen, Tel.: 0201 84735-0 www.unperfekthaus.de , Mail: info@unperfekthaus.de