INTERRUPTIONS #13
The inhuman voice

Curated by Genís Segarra

There is a long history of mankind’s attempts to build a machine capable of reproducing human speech. Some of the inventors who embarked on this quest where driven by curiosity – speech therapists and linguists interested for scientific purposes, for example –, while others were entrepreneurs with an eye to business opportunities.

The first talking machines date from the late eighteenth century, and many theoretical advances were made during the nineteenth century. But the turning point came with the emergence of electronics in the twentieth century. You can hear an example at 20’35” of this selection: a demonstration of the Voder (Voice Operator Demonstrator) at the 1939 New York World’s Fair.

The arrival of computers and microchips led to speech synthesis machines being marketed by companies like Bell Systems, Votrax, General Instrument, IBM and SAM, who developed them with the aim of replacing human beings in communications. At 27’38” you can hear the first computer that ordered a pizza by phone. ‘Domino? I want to order a pizza, a large pizza, pepperoni and mushrooms’, the machine says. Although it is fair to point out that the experiment failed, given that the Domino employee hung up on the computer. At 31’17” you can hear the first videogame that included a synthesised voice: an arcade shoot ’em up called Stratovox.

The mix includes several examples of talking software and microchips, but I’ve also thrown in songs that have used similar technology creatively: from German group Kraftwerk to the Japanese phenomenon of virtual singers. You will also hear songs that use a vocoder, an instrument that does not generate a human voice but can analyse the harmonics of a voice and then modulate it in another sound. This means that it can make any source of sound ‘talk’ or ‘sing’.

The vocoder was invented with the same aim in mind: to synthesise the human voice. Although it has now been superseded by chips that can generate vowels and consonants, artists and musicians have developed and used the vocoder in order to stand in for human beings. One of the first machines that achieved this effect was the Sonovox, which Disney used in 1941 as the voice of Casey Jr., the train engine in Dumbo.

In this mix you can hear Casey’s cheery ‘All aboard!’ at 17’01” and listen to him chant ‘I think I can’ as he struggles to climb uphill at 27’01”. The Sonovox was first used on a record in 1947, in the children’s book Sparky’s Magic Piano, in which a little boy discovers that his piano can talk and play itself. The voice of the piano was created with a Sonovox that transformed piano notes into a human voice. At 13’59” you can hear the fragment in which Sparky discovers that his piano can talk.

At the other extreme in terms of time and technology, the situation is much the same: at 13’18” you can hear a grand piano being ‘played’ by a computer-controlled mechanical system which manages to make the piano recite the Declaration of the International Environmental Criminal Court, a work created by the composer Peter Ablinger with the help of a software programme that assigns vowels and consonants to different combinations of piano keys.

Throughout the mix, you will hear vocoders and computers talking and singing. I’ve included several examples in which I’ve used vocoders or speech synthesisers in my own works with the groups Astrud and Hidrogenesse. There are also samples taken from a voice synthesiser competition held at the 2007 INTERSPEECH Conferences, in which participants had to make their programmes sing ‘The Synthesizer Song’. Several universities and companies participated in the competition and demonstrated their systems.

Genís Segarra, 2013