100% client-side
browser only
no server
no uploads
Everything stays on your device.
Your recordings are processed entirely in the browser using the Web
Audio API. Nothing is sent to any server — there is no backend, no
cloud storage, and no account required.
Clips are stored in your browser's IndexedDB and
persist across sessions. Use the download button on each clip to save
a permanent copy outside the browser.
this message only shows once
TURA.ops
record · reverse · play
Record
Speak
0:00
Hold button to record. Double-tap to lock · press again to stop.
Play
Hear reversed
no clip loaded
Waiting for audio…
Playback speed1.0×
Reverse Speech Analysisno model loaded
↻
Feeds your reversed audio directly into Whisper —
no re-reversing. The backward clip is the input. Whisper reports
any words or phonemes it picks out of backward speech, surfacing
sounds your ear might miss.
.en
English-only models are trained exclusively on
English speech. Faster to run, slightly smaller download, and more
accurate for English — all the model's capacity is focused on one
language. The hidden words it finds will be drawn from the English
phonetic space.
Best if: your recordings are in English.
multi
Multilingual models cover 99 languages and can
auto-detect the language — or you can give it a hint below.
Includes Macedonian, Croatian, Dutch, German, Serbian, French and
many more. Slightly larger download; marginally less accurate than
.en for English-only use.
Best if you want to explore any non-English language.