Synthesizers are controlled predominantly by keyboards. The reasons for this have to do with the historical acceptance of the keyboard as a practical and musically interesting man-machine interface. The disadvantage of relying on this means of control is that it limits the performer's abilities on a machine that was originated to remove limits. The keyboard is also a finalized type of device that has acquired a certain prejudice on how one should play it. These playing procedures are apt to ignore some potent keyboard — synthesizer information exchange possibilities.

The search for alternative and supplemental means of synthesizer control has involved many interesting areas, including brain waves. But the hands, feet and mouth are the most important areas of consideration. A look at the various non-electronic instruments controlled by mouth will point out the importance of this area. Note also that the hands are nearly always used in conjunction with the mouth and that examples of mouth only musical instrument control are hard to find (harmonica).

This brings us to a problem that synthesizers have in common; a lack of subtlety in their sound. This is due, in part, to the relative lack of synthesizer virtuosos, and in part to a reliance on various automatic circuitry to provide complex sounds. The "human" quality of music that is so important to its success and enjoyment is best delivered by the performer in a real time interactive process with his instrument. The subtle changes that a sensitive musician can introduce into the various parameters of his sound can have a powerful effect on the ear-mind of the listener. This effect is often most successful when the change is so slight as to be unidentifiable on a conscious level.

The mouth, tongue, throat and associated area are a particularly versatile source of sounds and can be controlled in very subtle ways by the attached organic computer. It seems obvious to take advantage of this to introduce information into the synthesizer, and thus realize a range of nuances that surpass non-electronic instruments.

Various approaches to utilizing the voice are gaining interest. The three most common involve amplitude to voltage conversion, frequency to voltage conversion, and vocoding. Examples of these techniques are appearing on records. (Vocoding — Herbie Hancock's Sunlight, Columbia JC34907; amplitude to voltage conversion — Patrick Gleeson's Star Wars, Mercury SRM-1-1178.) It is obvious from these and other examples that the vocal interface has great potential. A special need is in the polyphonic synthesizer, to replace hand control of pitch bending, etc., in the event both hands are being used on the keyboard.

CIRCUIT DETAILS

This particular design is a frequency to voltage converter using readily available parts. It is not expensive, but works quite well and will allow you to explore the possibilities of voice control.

Referring to Figure 1, IC1 is a BIFET op-amp configured as a mike preamp for an inexpensive dynamic element. The differential setup allows good hum rejection. The 1K resistors should be well matched — use 1% if you have them or find a close pair using an ohmmeter. The 1M feedback resistor determines gain. In this application, we want to avoid significant output from external sounds other than the performer, such as in a loud concert situation. In practice, you have to be very close to the mike to cause an output great enough to trigger the next stage.

IC2 is a Schmitt trigger-comparator with the trimmer determining trip level. IC4A serves to improve the rise time of the square wave output of IC2 for better results with IC3.

IC3 is a unique CMOS phase lock loop. Other PLLs do not have its capabilities and will not work in this application. In operation, the PLL locks onto the input frequency by detecting the phase difference between the outside signal and its internal VCO. It corrects any difference by changing the input voltage to the VCO. The 0.1 uF capacitor at pins 6 and 7, and the resistors at pins 11 and 12 control the frequency range of the VCO. For this use, the range is about 50Hz to 3KHz. Pin 10 supplies a voltage linearly proportional to input frequency. The 50K pot gives control over the maximum output level. With a 15 volt supply, about 12 volts is maximum. For some control inputs, you may wish to restrict this range using the pot.

The 1M Delay pot controls the charging rate of the 1.0 uF tantalum sample and hold capacitor, which is buffered by BIFET op-amp IC5. IC6 is a 4016 quad analog switch. One section, controlled by IC4D, switches rapidly on and off at the input frequency when a signal is present, and samples the voltage from the PLL. The other section dumps the sampled voltage to ground when activated. This provides a quick means to turn off the CV appearing at CV Out. The sample and hold function is a necessary addition to this design to avoid the control voltage sweeping down to zero in the absence of an input signal. Various settings of the Delay pot give the performer step or sweep functions and thus better control over the output voltage.

IC4B is a CMOS 4011 NAND gate configured in a linear amplifier mode. This provides sufficient signal level to activate one-shot IC4C and produce a trigger pulse. The 100K trimmer controls the trip level of the one-shot. This can be adjusted so that only a strong signal at the microphone will produce a trigger. Changes in the control voltage should be easy to produce without retriggering.

CONSTRUCTION AND OPERATION

Various prototyping boards are the quickest way to assemble this device; either temporarily on one of the plastic protoboards or on a regular PC board type for permanent use. Be sure to use a few .05 uF ceramic capacitors to bypass the power supplies. A couple of 10 uF electrolytics where the power supplies enter the board are also a good idea. The power supplies can range from +9 to +15 volts.

All of the ICs are static sensitive and sockets are a good idea. In any case, use a grounded or isolated soldering iron and avoid static prone clothing. If you're just a bit careful, it is very hard to zap an IC.

The TL071 op-amps have several substitutes. The RCA 3140, National LM356, or Texas Instruments TL081 will all work. The advantage of the TL071 is low noise.

Some thought should be given as to where you want the microphone. I have a large upright module cabinet on top of my keyboard. Figure 2 shows the setup. With the mike placed at mouth level, it is a simple matter to lean a few inches to use it. Of course, if you are using synthesis modules with a guitar, mounting the mike on a regular stand might be a better idea. In case you are wondering how to ground a mike element having only two connections, this is for the metal part of the mike (if there is one and it is isolated from the mike outputs), and/or the shield of a connecting cable. If the run is only a few inches, shielded cable is not necessary.

After construction, be sure to check everything before applying power. Proper orientation of ICs, electrolytic capacitors, diodes and power supplies is vital. Apply power and patch the CV out to a voltage controlled module with a steady signal input. Use a VCF if possible. Sing a steady tone into the mike and observe the Lock LED lighting. If it fails to light, adjust the IC2 trimmer. Adjust this for the most steady lighting of the LED at low frequencies. At one extreme, IC2 will be on constantly and will lock the sample and hold closed, resulting in the control voltage sweeping to zero.

Adjust the Delay control and observe the increased sweep effect as you sing higher and lower.

Connect the Trig Out to a module and adjust the 100K trimmer for an output only with a strong blast of air at the mike.

Attach the RST In to an appropriate source (or touch the input with your finger) and observe the control voltage jumping to zero.

USING THE MODULE

Although useful results can be obtained almost immediately, some self education in the skill of singing (and in using the voice in general) will provide a greater variety of effects. Don't limit yourself to regular types of sounds. Experiment with anything and have fun noticing what happens. Chances are no one will be able to hear what you are singing directly. This is great for anyone who is a bit hesitant about their vocal abilities. The previously mentioned records can provide some ideas as to types of effects that can be produced. In addition, the work of people such as Urszula Dudziak, who have experimented with synthesizer processed vocals, can be valuable input.

As an example of non-keyboard use, try triggering a sequencer and controlling a flanger of filter on the output of the VCOs controlled by the sequencer. Depending on the application, different levels and methods of introducing the control voltage may be tried. For example, when pitch bending or FM modulating a VCO, introduce a fairly small amount of control voltage via a 0.1 uF capacitor, for a typical effect. The accuracy of the converter is typically 1% over short ranges. This is not really as accurate as we would like for direct VCO control, especially if we are using multiple oscillators, but should pose few problems for other uses.

In the future, two additions to this module will be covered; an amplitude to voltage converter and a 4 channel control voltage switching and level control function. In the meantime, I'm sure you'll have fun exploring voice control.