Empaθi’s artificial singer(s)

Empaθi sometimes uses an artificial voice system, named Vox.

the Vox artificial voice machine

The Vox artificial voice

The Vox system produces a simulation of the human voice by digital means. More specifically, it models articulations of human vocal tract geometry. This is different from the more common method of concatenating pre-recorded syllables. No animals are harmed in the process.

The articulatory method enables very natural phoneme transitions, and subtle, individualized phonation & pronunciation effects. It is possible to produce the vowels for almost any natural language, as well as common consonants, with more exotic consonants in development. Suffice to say that dozens of languages and dialects are already available for Vox.

Vox can also adjust vocal register effects, just as we may use our modal or falsetto registers depending on pitch. Special effects such as growl and scream are in development. Vocal registers apply especially to the singing voice, and it is on this, rather than natural speech, that current developments are focused. But once we are happy with the overall design and workflow of Vox, we shall endeavour to return to modeling speech, especially to assist with hearing rare languages such as Sgai.

Features

flexible & dynamic phonation (vocal timbre)
flexible & dynamic speed of delivery
flexible pronunciation & expressivity
a natural presentation of an artificial voice, without “pretending to be” a natural voice
a humanesque voice, in the sense of a selectable digital instrument

State of the system

At mid-January 2017, the sound of Vox could have been clearer; it was still very suitable for trialing or rehearsing vocal parts. It was projected that as the voice quality kept improving, voice characters built with Vox should become independent soloists of some note.

New work was begun in 2019 to improve overall singing timbre and to model some missing phonemes. Strangely, the main male voice has settled into a very recognisable personality. The female voice is still proving a great challenge, however.

In January 2022, work has resumed with the help of hondrd. Many improvements in code efficiency and overall sound design have been made. Hard to believe this part-time project began eight years ago!

Some Vox sessions

1. A plain version of The Angel Gabriel, soon to be arranged in prog-rock style by Empaθi. Session September 2019.

2. Another version of Solomon’s Canticum V (see below). Session September 2019.

3. An extract from Empaθi’s new single Auto Pluto (of July 2019). Note the in-breath at 0'05" and the soft falsetto at the end.

4. An extract from a setting by Ian James of Solomon’s Canticum V – “ego dormio et cor meum vigilat vox dilecti mei pulsantis”. This has three voices in total, with a bit of reverb. Session January 2017.

(5). As an aside, you may also enjoy Empaθi’s instrumental track Pun (from November 2016).

Hiring of the Vox’s

Soon it will be possible to engage the services of Vox characters for participation in musical projects external to those of Empaθi. Enquiries may be sent to vox@skyknowledge.com.