Arabic kids story posted to Telegram with image and audio

Arabic kids story posted to Telegram with image and audio

An n8n workflow that posts a new children's story in Arabic to a Telegram chat every 12 hours. One model writes a short story in English, and a second model translates it. The Arabic text goes to the chat, followed by an audio version and a picture drawn from the story.

What is inside

• workflow.json: the workflow as n8n exports it, 15 nodes. Thirteen of them run and two are sticky notes. It holds no credential.
• README.md: these notes.

What we changed from the original copy

• OpenAI Chat Model and OpenAI Chat Model1 used gpt-4-turbo. OpenAI's deprecation page, read on 12 November 2026, lists it as shut down on 23 October 2026 and names gpt-5.6-sol. Both nodes use gpt-5.6-sol now, on version 1.2 of the node, with the model chosen from the list.
• OpenAI Chat Model2 named no model, so n8n would use gpt-5-mini. The page shuts that model down on 11 December 2026 and names gpt-5.6-terra. The node names gpt-5.6-terra.
• Generate an Image for the Story was an OpenAI node on version 1.3, which asks for DALL-E 3 unless told otherwise. The page lists dall-e-3 as shut down on 12 May 2026. The models it recommends in its place are themselves being shut down, and their rows name gpt-image-2.6-flare and gpt-image-2.6-sunburst. The node is on version 2.3 now and names gpt-image-2.6-flare. It still returns the picture as a file, which Send Image to the Channel posts.
• Create a Prompt for DALL-E is now Create a Prompt for the Image, because no DALL-E model is left in the chain.
• The three Telegram nodes held a chat ID from the original copy. It is the placeholder <YOUR_TELEGRAM_CHAT_ID> now.
• The large sticky note was a long description with a link to a channel. It is now a short note about what the workflow does and what to fill in.
• The credential links are gone, and the file's name is ours.
• Nothing else. The prompts, the text splitter, the speech step and the three posts are as the file had them.

Generate Audio for the Story is unchanged and keeps working until its model shuts down. OpenAI's speech guide, read on 12 November 2026, shuts tts-1 down on 6 January 2027 and says its replacement, gpt-realtime-2.1-mini, is not a supported model for the speech endpoint. After that date, swap that node for another speech service.

Mindket read this copy against n8n's source. It was not run on a live n8n and was not connected to a real Telegram chat.

Before you start

• An n8n instance, on n8n Cloud or your own server.
• An OpenAI API key.
• A Telegram bot and its token, and the bot allowed to post in the chat you choose.
• The ID of that chat.

Set it up

1. In n8n, create a workflow, open its menu and choose Import from File. Pick workflow.json.
2. Add a Telegram credential with your bot's token and select it in Send the Story To Channel, Send Audio to the Channel and Send Image to the Channel.
3. In the same three nodes, replace <YOUR_TELEGRAM_CHAT_ID> with your chat's ID.
4. Add an OpenAI credential and select it in OpenAI Chat Model, OpenAI Chat Model1, OpenAI Chat Model2, Generate Audio for the Story and Generate an Image for the Story.
5. Press Test workflow and check the chat for the text, the audio and the picture.
6. Publish the workflow, so it runs on its schedule.

What happens in a run

1. Schedule Trigger starts the run every 12 hours.
2. Create a Story for Kids asks the first model for a short tale for children of about 900 characters. It reads the schedule trigger's output as its input and has a Recursive Character Text Splitter of 500 characters with an overlap of 300 connected.
3. Translate the Story to Arabic asks the second model to put the story into Arabic with easy words and a moral lesson.
4. Send the Story To Channel posts the Arabic text. Generate Audio for the Story reads the same text aloud with n8n's default voice, and Send Audio to the Channel posts the audio with an Arabic caption that means "the end of the story".
5. In parallel, Create a Prompt for the Image asks the third model to describe the characters and how they look, with no text in the picture. Generate an Image for the Story adds an instruction to leave text out and draws it. Send Image to the Channel posts the result.

What to change

• The story's style and length: the prompt in Create a Story for Kids.
• The language: the prompt in Translate the Story to Arabic.
• The picture: the prompts in Create a Prompt for the Image and Generate an Image for the Story.
• How often it posts: Schedule Trigger.
• The voice: add a voice option to Generate Audio for the Story.
• Which chat: the three Telegram nodes.

Good to know

• The stories go to the chat as soon as they are made. Nobody reads them first, so read a few before you point the workflow at children.
• The story prompt gets only the trigger's date and time data as input, so the model picks the theme itself. Nothing in the workflow steers it.
• OpenAI's speech guide says your listeners must be told that the voice is AI-generated. Say so in the chat's description or in the caption.
• Each run makes three model calls, one speech call and one image call, and it runs twice a day. OpenAI bills each of them.
• The Telegram nodes send the text in Markdown unless you set a parse mode, so a story with an unpaired _ or * may be refused.
• Data: the story text goes to OpenAI, and the results go to Telegram.