Style training and voice profile
Automatic "my voice" training, training a style via @username, and how your voice fingerprint is measured.
last updated: June 11, 2026
my voice — automatic and free
When you connect your X account, the system scans your recent tweets and learns your writing habits. This is the "my voice" style: it doesn't count against any plan, costs no credits, and is used by default in generations.
Train with other voices
Give the @username of a writer you like — their recent tweets are scanned, their rhythm and structure are learned; you select that style when generating. You don't copy tweets one by one.
- +The style limit depends on your plan: Free 0 · Starter 3 · Pro 10 · Power 25 (my voice doesn't count against the limit).
- +If you select "all my styles (automatic)" when generating, the system blends the style best suited to the topic on its own.
Voice profile: a measured fingerprint
The Voice page shows characteristics ACTUALLY measured from your tweets: your rate of starting with lowercase, your use of questions, your emoji frequency, your average length... These aren't guesses but frequency measurements, and loom's generations are calibrated to this fingerprint — if loom drifts from your voice (for example, uses more questions than you), a corrective instruction kicks in on the next generation.