Crimean Tatar: Sevil female servinosmanov/speecht5-crh-sevil. Local SpeechT5 / Sevil female synthetic voices.
AI listening preview — native review pending.
Each600seconds,12 optional4-second inhale/6-second exhale rounds;final60seconds quiet. No breath holds;normal breathing at own pace whenever needed. General relaxation,not clinical advice or guaranteed sleep.
Written meaning and automated speech screens are limited and do not replace native pronunciation/naturalness review.
Voice model and cleaned Sevil dataset: Servin Osmanov; original Sevil dataset: speech-uk. Model Apache 2.0; dataset CC BY 4.0; Microsoft SpeechT5/vocoder MIT. Original generated narration modified by Happilogist, no model weights or source recordings bundled. Sources and attribution: see evidence/rights_review.json. Checked2026-10-19. No native certification or endorsement.
All previous archive payloads preserved unchanged.
