From an Arabic voice brief to a narration script: transcribe, verify, then rewrite

A voice brief is not a finished script

A merchant sends an agency a short recording about collecting an order from a branch. It contains a process to explain, an unconfirmed preparation time and a direction for the narrator. Pasting the whole transcript into text-to-speech would read out all three.

The text below is an original fictional brief for this walkthrough, not an actual client recording, a transcription result or a claim about a real shop. Its purpose is to show what a producer should resolve before narration.

Illustrative spoken brief: «نحتاج فيديو يشرح الاستلام من الفرع. العميل يختار الفرع وقت الطلب. يمكن يجهز خلال ساعتين، بس لا تذكرون المدة لين نتأكد. قولوا له ينتظر رسالة الجاهزية. وخلوه قصير وبنبرة هادئة.»

Meaning: explain branch collection and waiting for its confirmation message. Do not promise two hours before verification; “short and calm” is a production direction.

1. Transcribe a suitable, authorised recording

Sign in to the transcription page and upload an Arabic MP3 that you are allowed to process. If the voice note is in another format, convert it to MP3 in your usual audio tool first; changing the filename extension is not conversion. Remove unrelated private details before uploading, and keep the original recording for review.

When the transcript is ready, listen to the recording while reading it. Check names, numbers, negation and corrections especially carefully. In this example, losing “do not mention the time” changes the instruction into a promise the merchant has not approved.

You can edit and save the transcript in your account, copy it or download TXT. Keep the checked transcript separate from the narration draft so later rewriting does not erase what the speaker actually said.

2. Sort the checked text before rewriting

In your project notes, separate facts, open questions and production directions. Transcription does not verify a shop’s process or approve its claims.

• Process to verify: the customer can select a branch during checkout, and a message confirms that the order is ready. Check both steps with the merchant or the current ordering flow before writing them as instructions.
• Unresolved detail: “maybe within two hours”. Leave the time out. Ask for a confirmed policy only if the video needs a timing claim; do not change “maybe” to a guarantee.
• Production direction: “short and calm”. Keep it beside the script for the producer and narrator. Do not paste it into the spoken text.

3. Write the spoken version after the facts are confirmed

For this fictional example, assume the merchant has confirmed branch selection at checkout and the ready-to-collect message. The agency can then propose this short Saudi-style narration:

«تبي تستلم طلبك من الفرع؟ اختر الفرع عند إتمام الطلب، وانتظر رسالة الجاهزية قبل ما تتوجّه له.»

Meaning: “Want to collect your order from the branch? Select the branch at checkout, and wait for the ready-to-collect message before going there.”

The script now gives a customer action and its sequence. It omits the unconfirmed two-hour promise and the producer’s performance notes. This is a human editorial rewrite, not an automatic summarisation, translation or script-approval feature in FasihTTS.

Have the merchant approve the exact words before generation. If a timing policy is later confirmed, revise the script deliberately and approve the new version; do not silently edit an already approved file.

4. Choose a voice, then generate only the narration

Listen to Omar’s existing Najdi Saudi demo below before deciding whether that voice suits the piece. It is a pre-recorded public sample with a different script, not a recording of the collection instructions above and not proof of how your new script will sound. Its price and delivery line are fictional shop copy, not a FasihTTS offer or a promise about your business.

Copy the approved narration into the generator yourself and confirm the dialect and voice selection. The transcription workspace can open TTS in a new tab, but it does not automatically turn the transcript into a finished voiceover or clone the original speaker.

Review the generated audio against the approved script and the video. Check the order of the actions, the branch wording and whether the delivery leaves enough time to follow the screen. Make visual edits in your usual editor; FasihTTS is not assembling this video.

Keep the source, script and final audio distinct

Keep three clearly named items together: the checked transcript, the approved narration script and the chosen audio take. The transcript records the source; the script records the approved message; the audio is the deliverable to review. If a correction arrives, you can see which one needs to change.

Free samples are for evaluation and non-commercial testing. For client production, review the current paid options and Terms, and count both transcription and narration against the shared balance.