802.11 Wireless Microphone

From Catholicpenguin

Summary

What if you had a low cost compact 802.11 wireless microphone (voice recorder) which was internet ready by default?

Such a system would have numerous uses in several fields.

In security, such a device could be useful for survielance, replacing costly CCTV type networks with low cost wireless. Improvements in WPA would make them secure without need to run wires. Obviously for critical infrastructure, sensitivity to jamming would be important, but for many applications, 802.11 could be sufficient.

Such system could also be useful in doorbells, baby monitors, etc.

Finally, the target application I had in mind while thinking about this was classroom lecture recording. Such a system could be compactly installed as part of each podium, making it trivial for professors or speakers to record their lectures if desired; with the push of a button the lecture could be recorded, stamped by room and date, and placed in a directory on the professor's network share.

Technical Details

Like most of my networked ideas, the system could be fundamentally based on the fonera wireless router, retailing ~ $40. These devices have an Atheros SOC with wireless, Ethernet, and a ~200 MHz ARM chip onboard. Plus, the very well developed OpenWRT embedded linux distribution can be used on them.

A very common sound interconnect is I2S, or I squared S, a subset of I2C for sound transfer between ADC and DAC devices. There are already experimental drivers in the OpenWRT distribution for using the GPIO lines as I2C interfaces, so in the simplest case, the only hardware that would be needed would be:

• an I2C ADC of some sort, preferably w/ a built in mic preamp.
  • ( probably lots of people make these )
• a cheap condenser microphone, of the kind that go in walkie talkies, voice recorders, etc. (more then sufficient for voice recording)
• an optional ability to plug in an external mic might be nice.

Possibly one could find a chip with existing linux audio drivers, further easing development. Otherwise, a kernel audio driver would need to be developed. ALSA drivers could be used, but just old fashioned linux sound (/dev/dsp) would probably be easier.

Alternately, the Atmel AVR series (and probably several other vendors as well, but I'm most familiar w/ the AVR's) have fairly high speed ADCs built into their small microcontrollers, with built in differential preamps, making them ideally suited for audio ADC's. They can output the data over the RS-232 serial port, and the fonera has a RS-232 transceiver exposed via debug pins, so no soldering would be required.

Reading and writing from serial in OpenWRT Linux is as easy as reading and writing from /dev/console, so audio drivers could be written in usermode, easing debugging.

The rest is just software. Probably the fon has enough power to do real-time MP# encoding, so the audio could be level-corrected on the fly, mp3 encoded, and then the rest is up to you. Ideas include:

• Streaming via RTSP or some other (multicast?) multimedia streaming protocol to a perhaps QuickTime or Windows Media Streaming Server.
• Uploading as a file to a chosen Samba share
• Emailing (ick) to a specified address after recording is over.