All Radar picks

X OPPORTUNITY RADAR / DISPATCH

The Singer Cannot Be Arrested. She Also Does Not Exist.

Charlotte de Witte’s new Farsi-language techno EP uses an AI vocalist as a censorship loophole. The state cannot arrest the singer; the music industry still has to answer who gets the microphone.

Information Overload
2 min readSources
AI-generated conceptual editorial illustration of a vintage microphone locked inside an architectural cage while cyan and magenta sound waves escape and form a translucent woman patterned with Persian-inspired geometry. Government buildings and speaker stacks loom behind her; an empty pool of light marks the absent human performer. No real person or event is depicted.
AI-generated conceptual editorial illustration; not documentary evidence.

Iran’s system can prosecute a woman for singing solo. A synthetic performer creates a bleak loophole: there is no body for the state to arrest.

Charlotte de Witte released the three-track Azadi EP through KNTXT on November 5. The official release page identifies NAVA as an AI singer and Farsi voice created by Iranian-born, UK-based artist Farbod Mehr. It says all proceeds go to Femena, which supports women human-rights defenders and feminist movements in Iran and across the SWANA region.[1][2]

A voice built around an absence

Femena has documented the human cost behind the concept. The organization says Iran has restricted women from singing solo publicly for more than four decades and has used arrests, prosecutions and prison sentences against women artists. In August, it reported that singer Hiva Seyfizadeh had received a four-year prison sentence after her 2025 arrest.[3][4]

NAVA cannot be detained, barred from a stage or forced to sign a confession. That makes the project a real creative workaround to state censorship. It does not make the absent human performer any less absent.

The loophole has a bill attached

The release material identifies Mehr as NAVA’s creator, but the sources reviewed for this article do not disclose the model, training method or any human vocal dataset. Those unknowns matter. A synthetic voice can widen a censored stage, and it can also occupy space denied to the women whose repression gives the work its political force.[1]

The EP sends its proceeds to Femena and puts a Farsi voice in a global techno release. Those are concrete choices. The next questions belong to the industry: whose voices trained the system, who receives credit and compensation, and whether synthetic performance creates room for human artists or becomes a cleaner substitute for them.

The censorship workaround is real. So is the empty human-shaped space beside it.

Does a synthetic voice widen the stage—or take the microphone from the women the project says it represents?

FOLLOW THE SOURCES.

Checked Nov 5, 2026. Numbered references point to the evidence behind factual passages. Commentary is our own.

  1. Azadi EP (opens in a new tab)Charlotte de Witte / BandcampBack to cited passage
  2. KNTXT releases (opens in a new tab)KNTXTBack to cited passage
  3. Songs of Defiance: Iranian Women Singers Face Crackdown (opens in a new tab)FemenaBack to cited passage
  4. Four Years in Prison for Singing: The Price of a Woman’s Voice in Iran (opens in a new tab)FemenaBack to cited passage
MusicArtificial intelligenceIranCensorshipCreative technologyWomen’s rightsKNTXTFemena
Back to the latest picks