Audio recognition
The uploaded media is decoded and turned into a preliminary transcript with approximate segment timings.
Ayaty converts a recitation into timed verse segments, matches normalized speech against the Quran, and retrieves the canonical fully diacritized Arabic text before video editing.
The workflow uses several checks to reduce unrelated words and protect verse order.
The uploaded media is decoded and turned into a preliminary transcript with approximate segment timings.
Normalized Arabic supports matching, while the displayed result comes from canonical fully diacritized Quran text.
Verses stay ordered by Surah and verse number, with confirmed Surah transitions handled as separate passages.
Opening phrases are handled separately rather than searched as ordinary verses. If the reciter says one or both, they appear at their spoken time.
When the reciter repeats a verse or passage, the repetition remains in the timeline and follows the audio instead of being removed as a duplicate.
AI accelerates the workflow, while quick listening remains important for difficult recordings.
Play the segment and use verse correction if the Surah, verse number or text does not match what you hear. Ayaty rechecks the segment against Quran data.
Confirm that each verse begins with the recitation and ends after its final word. The timeline lets you adjust boundaries before export.
No speech-recognition tool can guarantee perfect accuracy for every recording. Noise, echo, reading pace and recording quality affect results, so Ayaty includes playback and correction before video creation.
Final verse captions are based on canonical Mushaf text. Unrelated speech should not be displayed, apart from separately handled opening phrases.
Sequence protection can recover an intervening verse when the surrounding matches make the passage context clear.
Yes. The upload page accepts common audio and video formats from iPhone, Android and desktop browsers.