Wireless VOIP?

zambuka asks: "Recently I was asked about pricing and availability of wireless microphones. What is needed are 15 wireless microphones, appropriate base units and the ability to record the signal from each microphone on a separate track. A session will last about 2 hours and the results need to be recorded to a CD. I have priced this at about $25000AUS probably less with bulk rates for the mikes and transmitters but still a big cost and not particularly portable for the base and recording units. The question I have regards alternatives. I have thought about using MiniDisc recorders, Nomad and Muvo. These are a much cheaper alternative and will suffice but require a major effort to transfer everything back to a PC and burn to a CD. I was also looking at wireless networks: 802.11a/b and Bluetooth and thought about how much cheaper and more suited to my needs this technology seems to be. Can these wireless systems cope with the constant throughput I need for this? And, is it possible (preferably with off the shelf components) to set up a wireless VOIP or similar network capable of dealing with this quantity of audio and recording the signals separately. Links to the tech would be welcome. Anyone living in Canberra who would like to take a shot at setting something like this up can feel free to contact me."

Comments

all sorts of problems

[TheSHAD0W]

I can see several problems with using 802.11b as a basis for a wireless microphone system, sufficient problems that I'd have to recommend you not attempt it, unless you really feel like tinkering.

First off, bandwidth may indeed be a problem, especially if you want high-fidelity results. A single mike recording at CD quality would require approximately 88K/s or 700 kilobits per second. 15 channels would over 10 megabits, assuming perfect efficiency -- which you wouldn't get with 802.11b without some fancy custom programming, so you'd need a set-up with at least 20 megabits of bandwidth. If you need to use these audio sources for professional mixing, the bitrate would double, to 40 megabits.

I'm aware of no commercial solution for a high-fidelity, low-power A/D converter for wireless microphone use. The ones I know about eat quite a bit of electricity and wouldn't be suitable for battery-powered applications. Building them would be difficult if you wanted to preserve fidelity.

You would further need to provide a small computer with each microphone and A/D converter, in order to feed that data onto the wireless channel. The task of coding for a single-chip processor or a small experimenter board would be difficult enough, but to keep electronic noise out of your audio circuit would be non-trivial. Powering it by battery would also cause problems.

Not quite enough information...

[stienman]

You've told us these specs:

• 15 wireless microphones
• Base units
• Record each signal on it's own 'track'
• Up to two hour recording period
• Record results to CD

From your further discussion, I'm assuming the following:

• Small microphone transmitters, user wearable, light
• Mobile users
• Small working area (wireless mics, 802.11 don't go very far)
• Medium or lower sound quality (good for voice, but not much else or you wouldn't suggest voip)

And this is enough information to suggest exactly what you suggested. If you told us better parameters, such as distance, typical usage (theatre, I'm assuming), obstacles between wireless microphones and recording system, etc.

I wouldn't suggest you do anything with voip right now unless you want to roll your own. It's more expensive, and at this point less mature and dependable than a basic wireless mic system. Unless you like babysitting the recording system...

So go with regular wireless mics. Be aware of contention and crosstalk. You can get wireless mic receivers that will support more than one mic simultaneously, and they are typically rack mountable. Put four 4 channel receivers in a small rack mount box with a rack mount computer. Now look for a multichannel recording card. You could go with 8 individual sound cards, but finding a mobo to support them would be a fun trick) The multichannel card should come with basic multitrack software. Perhaps a programs such as audacity will work with them. Beyond that you should be able to recording the resulting wav files to CD.

I wouldn't bother with having a bunch of little minidiscs of other individual recorders. To many little points of failure.

You might be interested in a few other options. You can get full duplex radios such as those used in cordless telephones from laipac.com. I'm currently using a cordless telephone for bike rides and other short distance communications. Last winter I mulled the idea over in my head of making a voip phone with 16 bit 802.11b cards (compactflash or pcmcia). It would take a relatively low power microcontroller to encode the communications. Since 802.11b doesn't have any real-time capabilites or QOS service, you won't get even packet rates. But since you didn't mention that the sound was going to listened to as it was created, and the computer can buffer just fine, you only need to time stamp each packet and the computer can still do a full recording with only a little buffer at the transmission side.

-Adam

Huh?

[adolf]

So. You need to record 15 things, without using too many wires.

Yeah, it'd be neat to use VOIP protocols like h323 and g711. But they're not robust, at all, by design. In intended use, if a network glitch trounces on a VOIP-transmitted spoken word (which happens fairly often, in a delicate balancing act between latency and fidelity), the other party might say "Could you repeat that?" In a recording enviroment, you have no such capability; the glitch is forever.

Additionally, there are no facilities that I'm aware of in common use which permit synchronization of multiple h323 endpoints. In other words, this means that the longer you're recording, the less well-synched things become (if you even manage to get them synchronous to begin with). After a couple of hours, you could well end up several seconds off.

Oh. And the codecs aren't so hot, either. Think 8KHz sampling rate, 64kbit/second maximum bitrate, and uLaw encoding. This is fine for telephone audio, but not for a $25k gig.

Ok, so it's not so neat to use h323.

And, don't get me started on 802.11/a/b/g. You're a fool to use this stuff for anything serious, as things currently stand - especially something serious with hard bandwidth requirements.

Why not do the old-skool thang? Buy/borrow/rent/steal a pile of decent wireless mics. Use whatever your local pro sound rental house suggests after you engage them in conversation about what, exactly, you're attempting to do here.

There's only a few considerations with wireless mics:

Make sure you can actually rack a lot of the recievers near eachother - oftentimes, the tuners go apeshit when in close proximity to eachother. If in doubt, try it beforehand.

You don't want dropouts, but you -really- don't want to hear the 500,000W AM station a few miles down the road.

And, make sure you're using the correct mic for the application. Since you're so fucking vague, we've got no idea if you're recording a press conference, a touring band, or the mating rituals of the Australian wolfhound. (Do you even want anyone's help?)

Oh. And many wireless recievers include some fashion of high-level output, suitable for driving a tape machine directly. You can avoid mixer rentals by using gear in this way, if dollars are tight. (OTOH, if possible, a good console and headphones to tie everything together will save numerous headaches.)

Speaking of tape machines, you need some kind of recorder. You can use a PC and a multi-channel pro-oriented sound card, but they're relatively unreliable - especially on the move (which you may or may not be - it's impossible to read through your obfuscation).

Instead, try multitrack tape. You've heard of it before. Modern nomenclature tends to sound something like "ADAT" when spoken. Such machines are common, cheap, and come in 8-channel chunks. Which means that you'll need two of them, and a toslink cable to keep their clocks synched.

Alas, ADAT is limited to 45 minutes or so per tape. This may limit your possibilities, or it may just mean that you need four machines in linked pairs in order to switch seamlessly to a fresh set of tapes.

Whatever the case, don't reinvent the fucking wheel for one gig. This is -just- analog audio here, and only 15 channels of it, at that. Don't make it any harder than it should be, especially if it's detrimental to the work (see paragraphs 2, 3, and 4.) Furthermore, if you're having a hard time understanding anything I've written here, find someone else to do this project for you; you're simply unqualified.