What is a multichannel recorder?

A conventional multichannel recorder is like stereo on steroids. Instead of just two channels, you have four or more. Each of the channels can be a completely independent sound track. But the most common use is to make each channel record a different "voice" in a musical performance. After the recording is complete, you can mix the channels together to create a final product (e.g. "Turn up the drummer a little, and lower the background singers.")

The chief problem with a traditional multichannel recorder is that they are very expensive. Decades ago, multchannel recorders were used only by professional recording studios, built to high standards, and priced accordingly. Rick's position in the entertainment industry gave him access to equipment beyond the reach of many home haunters.

Since then, prices have dropped somewhat as the market has expanded - because garage bands need equipment, too. And technology has progressed, notably in the areas of computerized audio. But we wanted something simple and cheap.

We aren't going to attempt a real multichannel recorder and player. It would be a big project and we really don't need true multichannel capability.

Instead, we intend to produce a similar "good enough" effect with a small investment.

Basic Concepts

• Performances are recorded on stereo media with two utterly different channels.
• The audio channel contains all of the audiable material - all the singers and instruments mixed together.
• The control channel indicates which of the performers "sing" the material.

The control channel contains "DTMF" tones. These are the noises made when you dial a telephone, and are sometimes called "touch tone". Just about all modern telephones use this tone dialing method. [Before that, telephones dialed using a series of pulses.]

Our controller will listen for the DTMF tones on one audio channel and use them to figure out which characters should be speaking. The controller then routes the audio signal only to the characters who should be performing.

This illustration shows how the the control channel switches the audio channel to the outputs:

We will use the following terminology:

• channel - Our preferred term for an information conduit, usually carrying electrical signals representing sound.
• track - A specific piece to be performed, e.g. track 1 is "Dead Man's Party", track 2 is "Come Little Children".
• right channel - One output of a conventional stereo player, feeding signals into this project. The right channel will carry presentation-quality audio.
• left channel - One output of a conventional stereo player, feeding signals into this project. The left channel does not contain anything you listen to.
• control channel - The channel that contains information about which performer(s) should sing.
• audio channel - The input channel that contains all of the audiable content.
• speaker audio - Audio channel good enough to be amplified and presented to audience through speakers.
• animation audio - An audio channel that is intended to be fed into mouth-moving mechanism. Because you aren't going to listen to it, "animation audio" may be low quality; optimized for animation mechanisms; or otherwise unpleasant to listen to.
• performer - A single singer (or a group of singers singing in unison).
• voice - An output channel, sent to a performer.

Refinement

Suppose that the material consists of characters singing, with instrumental accompanyment. At the end of the song, all the singers hold their last note, while the big bass drum beats "boom, boom, boom!" Lovely! But we have mixed everything down to mono, so the only music we hear is a combination of everything, bass drum and all. Our control channel merely allows us to choose which performers will make that big combined sound. So when we approach the final note, the control channel says "everybody sing", and the combined sound track comes out of everybody's mouths - and for each "boom" of the drum, their mouths open wider.

What we really need is three channels:

• An "animation audio" channel that moves the singers' mouths. It should have everything except the instruments. Think of it as "reverse karaoke" channel. If you don't have the equipment to extract a reverse karaoke channel from an existing recording, you can record your own channel just singing along - After all, this is fed into the motors; nobody will hear it.
• A "speaker audio" channel that is suitable for amplification and presentation to the audience.
• The "control channel" that indicates who is singing now.

But all we have is stereo - two physical channels. We are embarking on this project because we don't have a third channel, or a fourth.

What we will do is combine the animation audio and control channels on a single physical recording channel (the stereo's left channel). But in order to combine the two channels, we need a way to split them apart when we play back. We have decided to use a specific tone frequency for the animation audio that can be filtered out. The outline is as follows:

• We create the animation audio channel by extracting the volume envelope of the vocals, using that to modulate a 3 kHz carrier.
• We mix that 3 kHz signal with the DTMF tones, placing both animation audio and control channels on a single physical recording channel.
• The 3 kHz carrier is well out of the way of the DTMF tones, the highest of which is 1633 Hz.
• By using simple low-pass and high-pass filters, we can recover the individual channels for animation audio and control.

Future

We have decided that the way to do this is to add a second tone above the DTMF sounds, perhaps 3.5 kHz. This channel will be distributed to all props, in the animation audio channel. This tone can be recovered by a simple circuit in each prop, and turned into a boolean on/off bit-stream.

The boolean bit-stream can be used to send commands using a serial protocol to be defined later.

We're not designing this now, because we don't need it now. But we wanted to have the flexibility to add it alter.

Related Pages