Convert Arabic audio to text

Modern Standard Arabic (news, formal speech, lectures) transcribes best. Widely-spoken dialects like Egyptian and Levantine work reasonably; Gulf and Maghrebi dialects are more variable. Clear audio helps every dialect.

Yes: proper Arabic text that pastes correctly into Word and websites. TXT, DOCX, SRT and VTT all handle RTL correctly.

Recitation (tajwid) has elongations and melodic patterns that speech models mis-handle; published transcripts of recitation are a better source. Everyday speech is where Trustample helps.

No. Output is standard unvocalized Arabic, the way newspapers and business documents are written; full tashkeel marking is a specialist editorial task that no mainstream speech model performs. Short vowels stay implied by context, exactly as in normal written Arabic, and where a word would be ambiguous without them you can add the marks manually in the editor.

Broadcast and lecture audio is the best case twice over: it is usually Modern Standard Arabic, and it is professionally recorded. A voice note is dialect through a phone microphone, so expect a rougher draft. Still worth uploading (save the note as an audio file and it uploads directly, OGG and M4A included); just plan a heavier inline pass.

Common borrowed words survive. Heavy sentence-level switching, the Arabic-French weave that runs through much North African speech, is genuinely hard for every model; transcribe under Arabic when Arabic dominates, or under French when the recording is mostly French with Arabic phrases.

Not currently. The model generation that supports Arabic doesn't accept our vocabulary boosting yet, so that box is ignored for Arabic uploads. Fix names inline in the editor instead; each correction is one click and carries into every export.