Join the microphone and audio lines from a headset for recording

Question

Question

What are the steps required to join the microphone and audio lines from a typical cell phone headset into one line so that the audio can be recorded by a standard voice recorder using the microphone and/or line-in input?

Practical use

The practical use for this is for the purpose of recording cell phone conversations with the use of a headset.

In Depth

Intuitively, I would just splice all the lines together. However, this approach seems to have the problem of creating an echo from the mic line to the audio line. I've also heard that it could blow some circuits in the phone, audio recorder or both.

It seems that the job of "splicing: these lines together without interfering with the call is more complex, requiring the use of some components to make it work correctly.

Here is a device that I found on Ebay that does exactly this:

I purchased a similar device off Ebay that has its own headset attached and cracked it open:

Here is a close up of the cracked open part. It does not look very complex, but I'm not sure exactly what it is doing:

(Other side:)

Answer 1

What are the steps required to join the microphone and audio lines from a typical cell phone headset into one line so that the audio can be recorded by a standard voice recorder using the microphone and/or line-in input?

In this context "audio line" means a signal level & output impedance intended to drive a typical pair of earphones. I'll assume mono rather than stereo.

I'd guess the processing steps may be not entirely unlike the following:

1. for mic-in (could be entirely passive?)

   1. Attenuate the headphone signal to mic levels
   2. mix audio signals
   3. buffer output

2. for line-in (choose either pro or consumer level)

   1. amplify the microphone signal to line-level.
   2. amplify or attenuate earphone signal to line-level
   3. mix audio signals
   4. buffer output

By buffer output I mean set the appropriate output impedance.

For stereo, you'd typically split the mic signal and mix equally into left and right channels.

For recording conversations you might want separate automatic gain control (AGC) for each side of the conversation. Since conversations are mostly half-duplex I guess you mostly only need one AGC element.