Machine Learning Poetry, the new LP from Tehran’s kāve, is not pastoral or pretty. It’s 10th-12 century poetry made into a brutal digital assault – machine learning sound that cuts like a full-force buzz saw. And it’s fantastic.
We’d expect nothing less from the sonic world of Ata ‘Sote’ Ebtekar, who runs the experimental Iranian label Zabte Sote. But at the same time, Machine Learning Poetry is an answer to dumbed-down grids on one hand or senseless complexity on the other. On repeated listenings, kāve’s album reveals itself as language, with nuance and shape, even if it feels slightly like you’ve been abducted out of a field by some not terribly gentle aliens into the mothership.
But also, crucially, this sounds very unlike other recent brutalist electronic sounds or the relative sameness of AI artifacts. The machine learning in this case is focused on the rhythms, and the rhythms in turn derived from Persian authors of the pre-classical period. The sounds are distorted and mangled voices, so not AI-generated timbres – which keeps kāve’s timbral voice in the fore. SOTE’s rework matches perfectly, and has all the label boss’ signature power.
Out today, here’s the accompanying description of what you’re hearing and how it connects to verse from around a millenium ago:
Machine verses; it regenerates rhythms and rhymes based on a robāʿī (quatrain) by Onṣorī (10th-11th century) and a qaṣīdeh (ode) by Khāqānī (12th century); two of the most iconic and influential Persian poets between 10th to 12th centuries.
The process includes:
Metadata (cadence, rhythm, rhyme, accent) analysis and recording verses. Making MIDI clips out of the poems.
Sound synthesis based on the recorded voice as the source material. Letting the machine learn poetry.
With kāve’s music as with Sote’s. somehow the result, violent as it is sonically, can be healing. It never feels forced; it’s powerful, but like a volcano – it seems to exist in the form in which it exists. (Kaveh studied with Ata once upon a time.)
And the artist Kaveh Sattari is one to watch – triple force musician, sound artist, and mastering engineer. He has engineered his own masters here, too, pushing the throttle all the way forward without losing a drop of clarity.
Five of the tracks have their own short companion videos by the artist as a kind of audiovisual calling card; here’s the full set:
Also check the intense propulsion of his live set at Tehran’s 2020 Set Festival (excerpt). I hope some day this is the kind of music we all dance to. It’s unquestionably dance music, a new evolution of industrial into something entirely unique – without the blandly mechanical approach to groove, but more organic and more powerful at once.