Introduction: Turn-Taking Device
Conversation analysts consider turns to be the most basic unit of conversation. As you would imagine, the intricacies of turn-taking have been well documented by researchers, and this process always involves the careful transcription of pre-recorded conversation.
I am no expert but wanted to develop a recording tool for my own personal research. Below you will see how I developed my first rudimentary prototype. In short, the device monitors who is talking in a 2-person dialogue and measures the length of each turn.
Step 1: Gather Materials
Below are the components I used:
Electret Microphone Amplifier (x2)
MicroSD card breakout board
SD/MicroSD Memory Card (4 GB SDHC)
USB MicroSD Card Reader/Writer
3.5mm (1/8in) Stereo TRS Male to Male Audio Cable (x2)
3.5mm Stereo Input Jack (x2)
2 small light bulbs
1 push button
2 mm screws
Step 2: Assemble Components
The components are assembled as shown in the above diagram. The sketch works in the following manner:
When the record button is pressed, the Arduino monitors the volume level from each mic. Theoretically, the louder mic indicates which person is speaking, and this is communicated by one of the two lights. For example, if mic A volume is higher, lightbulb A turns on, indicating that person A is speaking. The length of each turn is measured using a simple timer and is written to a text file on the SD card. Pressing the record button a second time stops the loop and closes the text file.
Step 3: 3D Print Case
The case is simple. It includes precise mounting places for each component. I used the Objet Vero White material for most of the case and later printed two rectangular pads in rubbery Tango Black to keep the mics from slipping around on the table.
Step 4: Assemble Case
I stripped one end of each audio cable and used the three inner wires to connect OUT, GND, and VCC from each mic to the Arduino. Once the components were all functioning correctly, I mounted everything inside the case with a combination of screws and super glue. Vero White is not the best material for screws but it works well enough for a prototype.
Step 5: Finished
While the prototype functions, it could use several improvements. The low cost mics I used are not the best for tracking volume levels, especially when close together. Higher quality, unidirectional microphones will help solve this issue. Also, a more durable version of the case will allow for a better mounting surface for the audio input jacks.