Voice editing software

Neural voice tools

Software for editing and transforming the voice: the neural tools I develop at Supertone and IRCAM.

Supertone’s pitch drawing tool

At Supertone I build neural voice-editing tools on top of our NANSY voice backbone. Presented at ISMIR 2025, the latest tool edits pitch, subharmonics and structural noise (jitter and shimmer): it decomposes a voice into pitch, linguistic and speaker representations, and you reshape it by redrawing the control-parameter curves in a web UI, on a computer or tablet.

The pitch drawing tool is available for business customers at Supertone and is used by audio engineers and movie directors.

IRCAM’s CIRCE

CIRCE — the IRCam neural auto-enCoder for voiCE — transforms the pitch and intensity of a voice using the neural auto-encoder developed in my thesis. Originally built to realise the voice transformations for the artistic residency of Judith Deschamps, I’ve refined the design in collaborations with composers such as Omer Barash, Sachie Kobayashi and Aïda Shirazi.

The original app was a GTK standalone. With the help of our software engineer Pierre Guillot, we’ve ported the model into a VST plugin. The voice transformation runs in real time on modern computer hardware.

CIRCE is available on the IRCAM forum to any registered user (registration is free). The research and software were developed as part of the ANR project ARS (ANR-19-CE38-0001-03).