I made an Instagram Reel with AI, from the script to my cloned voice and animation

Brainstorm the idea, then write it for the voice

Start a conversation with Claude about the ideas you could explain to your audience. Discuss the angles until you choose one, then ask it to write the script in your speaking style. Give it a sample of your language so the result sounds like something you would actually say. Before copying the text into ElevenLabs, ask Claude to prepare it with delivery annotations appropriate for the voice model you will use. Tell it where you want a quiet delivery, a stronger voice or silence. Read the annotated version and listen to how the model interprets it; exact delivery and pause timing may need another attempt.

Create your voice and choose the take you want

Create your own voice in ElevenLabs using recordings of yourself. The almost 45 minutes I mentioned were recordings used to train my voice clone, not 45 minutes spent making this Reel. Then paste the annotated script and generate the narration. I used Eleven v4. My first generation returned two variations. I listened to both and chose one to give to ChatGPT. Listen for the delivery you asked for and the way the words sound in your dialect. If neither variation works for you, change the wording or annotations and try again. Download the chosen audio; that is the recording you will use for the video.

Brief ChatGPT on the design, the hook and the 60-second limit

Upload the chosen voiceover to ChatGPT in a setup that can produce and render video. I used ChatGPT 6.1 Sol. My own production skill is still being prepared. I will add its GitHub link here when it is ready so you can give it to ChatGPT with the audio. Ask for motion graphics and original design decisions across shapes, corners and colors. Give it visual references such as my Memory Engine page and mentorship page. Explain that the opening must capture attention in the first 3 seconds and the finished Reel should last 60 seconds. Allow it to reorder the script, cut lines or silence, and slightly speed up the audio when useful. If it proposes new spoken wording, have the replacement audio generated in your cloned voice.

Keep iterating, then save what you agreed

Watch the version ChatGPT produces and keep giving it specific feedback until it matches what you want. Say which opening feels slow, which visual looks generic or which transition needs changing. The production prompt below is a starting brief; the finished style comes from those decisions and revisions. Once you approve the result, ask ChatGPT to record every agreed edit and preference in a reusable brief or the project instructions, and confirm where it saved them. Keep an accessible copy and ask it to read that record when you start the next Reel. This gives the next attempt a better starting point; it does not guarantee a final result on the first try. Watch the exported video before posting it.

كيف تعمل ريل بصوتك بمساعدة الذكاء الاصطناعي

هالريل بدأ من جلسة عصف ذهني مع ⁨Claude⁩. ناقشنا الفكرة، وبعد ما اخترتها طلبت السكربت مع تعليمات الإلقاء المناسبة لـ ⁨ElevenLabs⁩: وين الصوت يهدأ، وين ترتفع النبرة، ووين نحتاج وقفة.

جهّزت نسختي الصوتية من قرابة ⁨45⁩ دقيقة من تسجيلاتي، وولّدت التعليق باستخدام ⁨Eleven v4⁩. أول توليد أعطاني نسختين. اخترت اللي عجبتني وأعطيت ملف الصوت لـ ⁨ChatGPT 6.1 Sol⁩ ليعمل الفيديو.

طلبي كان واضح: ريل مدته ⁨60⁩ ثانية، بتصميم أصلي وموشن جرافيك، وبداية تشد الانتباه من أول ⁨3⁩ ثواني. وأعطيته حرية يغيّر ترتيب الكلام أو يختصره إذا هالشي بيخدم الفيديو. بعدها ظلّيت أطلب تعديلات لحد ما وصلت للنتيجة اللي بدي ياها.

آخر خطوة كانت حفظ القرارات والتعديلات حتى ما نبدأ من الصفر بالريل الجاي. عم جهّز ⁨skill⁩ خاصة بهالطريقة، ورح أضيف رابطها على ⁨GitHub⁩ هون لما تجهز. حالياً، فيك تمشي على خطوات الإنتاج الموجودة هون.

تابع حسابي على الانستغرام:

رابط الانضمام للدورة: https://halaqa.app/enrollment?course=start-with-ai

الخطوات

Prompt

# Attach the chosen voiceover and script, then paste this production brief.
# Create a 60-second vertical Instagram Reel with original motion graphics.
# Build an opening that captures attention within the first 3 seconds.
# Use creative shapes, corners, colors and compositions; avoid generic AI design.
# For visual inspiration, study these pages:
# https://www.aiwithmo.com/lp/memory-engine
# https://www.aiwithmo.com/mentorship
# Adapt the visual direction to the message instead of copying a page.
# Reorder the spoken segments, trim lines or silence, or slightly increase
# playback speed if needed to reach 60 seconds while keeping speech natural.
# Suggest replacements where they improve the hook or explanation; flag any
# new wording that needs a replacement recording in my cloned voice.
# Use the available production tools to make the video, with sound effects
# that support the motion and leave the narration clear.
# Show me the result so we can revise it together. Deliver the finished export.
# After I approve it, save all agreed edits and design preferences in a reusable
# project brief, tell me where it is saved, and read it before the next Reel.