Arabic Voice Dictation Skill
/install arabic-voice-dictation-skill
Arabic Voice Dictation Skill
Purpose
Use this skill when the user provides Arabic voice notes, mixed Arabic/English speech, or rough dictated notes and wants a clean transcript, summary, task list, or bilingual copy without losing meaning.
The workflow is optimized for Arabic-first use cases: Gulf dialect notes, Modern Standard Arabic polishing, right-to-left formatting, Arabic/English mixed terminology, and safe handling of personal or business information.
When to use
Use when the user asks to:
- Transcribe Arabic voice notes or audio they are authorized to process.
- Convert rough Arabic speech into clean MSA, Kuwaiti/Gulf Arabic, or bilingual Arabic/English text.
- Extract decisions, reminders, tasks, shopping lists, meeting notes, or content ideas from Arabic dictation.
- Prepare WhatsApp, Telegram, email, social caption, or report drafts from dictated Arabic notes.
- Check Arabic grammar, tone, RTL layout, names, numbers, dates, and mixed English technical terms.
Do not use this skill for covert recording, unknown third-party audio, secrets, passwords, legal evidence handling, or identity documents unless the user confirms authorization and the output is limited to safe summarization.
Inputs to request
Ask only for the minimum needed:
- Audio file, transcript text, or rough dictated notes.
- Desired output: transcript, summary, action list, cleaned Arabic, English translation, or bilingual version.
- Dialect preference: Gulf/Kuwaiti, MSA, simple Arabic, or keep original dialect.
- Tone and destination: personal note, client message, social post, report, study note, or task list.
If the audio/transcript contains names, phone numbers, addresses, client details, or private account data, preserve only what is necessary and avoid repeating sensitive details in chat unless the user explicitly needs them.
Workflow
-
Confirm authorization and purpose.
- The user must own the audio/notes or have permission to process them.
- If the source is unclear, ask one short authorization question before continuing.
-
Transcribe or parse the note.
- Keep Arabic text in correct RTL order.
- Preserve names, numbers, timestamps, and important English terms.
- Mark unclear words as
[unclear]instead of guessing.
-
Normalize safely.
- Fix obvious speech-to-text errors.
- Keep meaning unchanged.
- Do not invent missing facts, promises, prices, dates, or client commitments.
- For dialect-to-MSA conversion, keep culturally natural phrasing.
-
Extract structure.
- Summary: 2-5 bullets.
- Action items: owner, task, deadline if stated.
- Questions: items that need follow-up.
- Draft copy: only if the user requested message/social/report text.
-
Produce the requested output.
- For Arabic-only requests, answer Arabic-only.
- For bilingual requests, show Arabic first, then English.
- For technical terms, keep accepted English words when Arabic translation would confuse the reader.
-
Verify before finalizing.
- Check RTL punctuation, Arabic numerals/dates, names, and line breaks.
- Label uncertain items clearly.
- Do not claim the transcript is perfect if audio quality was weak.
Output templates
Clean Arabic note
العنوان: [short title]
الملخص:
- ...
- ...
المهام:
1. [task] — [owner/deadline if stated]
2. ...
نقاط تحتاج تأكيد:
- ...
Bilingual summary
Arabic:
- ...
- ...
English:
- ...
- ...
Action items:
1. ...
WhatsApp-ready message
[Greeting]
[Main point in 1-3 short lines]
[Question or next step]
Social/content idea from voice note
Hook:
...
Main points:
1. ...
2. ...
3. ...
Caption draft:
...
Hashtags:
#... #... #...
Quality checklist
Before replying, verify:
- Arabic reads naturally and is not literal machine translation.
- RTL punctuation and line order are clean.
- Names/numbers/dates match the source.
- Dialect/MSA preference is followed.
- Unclear words are marked, not invented.
- Private details are minimized.
- The answer includes a clear next step if confirmation is needed.
Safety rules
- Do not process audio that appears secretly recorded or unauthorized.
- Do not ask for passwords, OTPs, private keys, recovery phrases, or session cookies.
- Do not expose unnecessary personal data in summaries.
- Do not create legal, medical, financial, or HR conclusions from a voice note; summarize and recommend professional review where needed.
- Do not impersonate someone from a dictated message.
- If publishing, sending, or uploading is requested, prepare the text only and ask for explicit approval before any external action.
Example prompts
- "Clean this Kuwaiti Arabic voice note into a short WhatsApp message."
- "Summarize this Arabic meeting note and give me action items in English."
- "Turn my Arabic dictation into a professional Instagram caption."
- "Keep the dialect, fix mistakes, and mark unclear words."
- "Make this Arabic note bilingual for a client handoff."
Support / Donate
If this skill helps your workflow, you can support maintenance here:
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install arabic-voice-dictation-skill - After installation, invoke the skill by name or use
/arabic-voice-dictation-skill - Provide required inputs per the skill's parameter spec and get structured output
What is Arabic Voice Dictation Skill?
Capture Arabic voice notes, transcribe them safely, clean dialect/MSA text, preserve meaning, and prepare bilingual summaries or action items. It is an AI Agent Skill for Claude Code / OpenClaw, with 35 downloads so far.
How do I install Arabic Voice Dictation Skill?
Run "/install arabic-voice-dictation-skill" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Arabic Voice Dictation Skill free?
Yes, Arabic Voice Dictation Skill is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Arabic Voice Dictation Skill support?
Arabic Voice Dictation Skill is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Arabic Voice Dictation Skill?
It is built and maintained by Kw.Hades- Creative Labs (@abdullah944); the current version is v0.1.0.