A fine musician complemented by a fine musical instrument can communicate a range of strong emotions. But even the best of these combinations cannot speak in the sense of transferring textual information. Which leads to the question: how might a player make a musical instrument “talk”?
One answer is to use a talk box, a device that allows the sound of a musical instrument (usually a guitar) to be controlled by the player’s mouth. Since the early 1970s this effect has been used by many musicians – perhaps the most famous is the guitarist Peter Frampton, particularly playing Do You Feel Like We Do or Show Me the Way.
The basic device usually consists of a compression driver (essentially a loudspeaker without a large vibrating cone) connected to a long and narrow flexible tube.
The other end of this tube is usually attached to a microphone stand.
It is arranged so that the open end can be inserted into the corner of the player’s mouth when that microphone is used. The output of the guitar amplifier can then be connected to the compression driver in the talk box instead of the normal loudspeakers. The sound of the guitar is thus injected into the player’s mouth.
To understand why this can be so effective, we need first to consider the voice. In both speech and singing, vibrations of our vocal folds (small flaps of tissue in our voice box located behind the Adam’s apple) periodically interrupt the airflow from the lungs.
Depending on the shape of the mouth, this sound is altered in different ways before it emerges from the lips: we say it is filtered by the acoustic resonances of the vocal tract. Moving our tongue, lips and jaw alters the magnitudes and frequencies of these resonances, and consequently the harmonic content of the voice.
Vowels are particularly important in singing, and they are largely determined by the two resonances at the lowest frequencies.
When a talk box is used, the player simply mimes the intended speech with the tube placed in the mouth. It is now the guitar sound, rich in harmonics, rather than the voice that is filtered by the vocal tract before it reaches the microphone.
The harmonic content of the guitar sound carries the acoustic information about the shape of the mouth so it sounds as if the guitar is talking.
Of course, the guitar input is very different from the input normally generated by the vocal folds, but this does not usually pose any difficulties in comprehension. This is because we are already used to understanding speech that has been produced with different sounds from the vocal folds, e.g. normal voice, falsetto voice, a croaky voice and whispering.
The technology used in the talk box is quite simple. It is thus interesting to consider why they are still used some 40 years later. One alternative might be the “wah-wah pedal”, an electronic device of a similar vintage that only adjusts a single resonance.
But this means that, unlike the talk box, it cannot adjust the multiple resonances that are required to mimic speech.
During those intervening 40 years there has also been immense progress in the speed and power of digital signal processing. It is consequently now possible to implement Vocoder software that operates in real time.
This is capable of replacing the detailed harmonic spectrum of one instrument with the harmonic spectrum of another instrument (usually, but not necessarily, the voice).
A talk box is different in that the sound is, and can only be, changed in real time. It also has the advantage that the substitution occurs inside the mouth and will be heard by the player in the same fashion as if they were actually speaking. This can make a difference – your recorded voice sounds different from that you hear yourself while speaking.
There is also an interesting parallel with research carried out in the Acoustics Lab at the University of New South Wales. We also inject a signal into the mouth from a tube located just outside the mouth, but instead of the sound from a guitar we use a carefully synthesised sound.
This sound contains a large number of frequency components, and this allows precise measurements of the resonance frequencies of the vocal tract.
Measuring the acoustics of the vocal tract in this fashion has allowed us to determine how singers adjust the resonances of their tract to match harmonics of the sung pitch.
That’s particularly important when singing at high pitch. It has also revealed how musicians can use their vocal tract while playing musical instruments such as the didgeridoo, saxophone and clarinet.
The simple concepts behind the talk box have thus not only allowed guitars to talk – they have also helped discover how singers and wind players use their vocal tracts in musical performance.