Home › Blog

How to Make a Hindi Story Video with AI Voices

By Tooljin · 6 October 2026 · 15 min read

Kahani channels are a big part of Hindi YouTube: bedtime stories, moral stories, horror, mystery, mythology. The hard part of making one has never been the story. It's everything around it. A narrator, a different voice for every character, a door creak at exactly the right moment, music that sits under the voices instead of over them, and pictures where the hero looks the same in scene one and scene twelve.

This guide walks through the whole thing in one tool, the Tooljin AI Voice Generator, using a real example you can listen to below. Nothing here needs editing software, a microphone or a voice artist. Your first video will take the longest, mostly because of the pictures. After that, most of your time goes into the part that matters: writing.

What you need

• A story, or just an idea. A two to three minute story is roughly 2,000 to 2,500 characters. Hindi in Devanagari, Hinglish in Roman script and English all work, and so do 22 more languages, from Tamil, Bengali and Marathi to Spanish and Japanese. No story yet? The tool can write one for you (see Step 1).
• A Tooljin account. The free plan gives you 200 credits a month (up to 50 a day in this tool), enough to try a story of about a minute. A full three-minute video needs a paid plan.
• An image tool you already use, such as Google Flow or ChatGPT, for the pictures. The voice generator writes the prompts; you make the images wherever you like.
• Chrome or Edge on a computer is fastest for the final video, and phones and tablets make it too, in Full HD; with Pro, a laptop or computer can also make it in 4K. You can write and make the audio on a phone.

Step 1: Write the story as a script

Open the voice generator and switch to Story mode. Story mode reads your script the way a director would: who speaks, how they say it, and what we hear in between. The format is simple:

• One line per speaker: Name: what they say. The narrator is just another name. Many people call theirs Sutradhar.
• How to say it goes in square brackets at the start of a line: [scared], [whispering], [softly], [laughs]. Brackets are directions. They are never read out.
• Effects get their own lines: Sound: darwaza for a single sound, Music: rahasya and Ambience: raat for music and background that keep playing.
• A blank line is a scene change and adds a longer pause. [pause] adds a short silence.

Here is a real script, exactly as it was typed into the tool:

Music: rahasya
Ambience: raat
Sutradhar: Raat ke barah baje, purani haveli ka darwaza apne aap khul gaya.
Sound: darwaza
Golu: [scared] Dadi... wahan koi hai kya?
Dadi: [softly] Darr mat beta. Ye toh bas hawa hai.
Sound: bijli
Golu: [whispering] Toh phir... ye lantern kisne jalaya?

And this is what the tool made from it, with no editing afterwards. Three character voices, two sounds and background music, in 33 seconds:

Two habits make a big difference. Keep each line to one breath and one emotion; a long line loses its feeling halfway through. And give each character a way of speaking that's different from the others, so a listener always knows who's talking even with their eyes closed.

Already have your story written as a paragraph? Paste it and press ✨ Format. The AI sorts it into speaker lines and adds delivery cues, without adding or changing a single word. If it changes even one word, the result is refused and you aren't charged. It costs 23 credits per 3,000 characters.

No story yet? At the top of the Story tab, the 🎬 Story Director writes one for you with the tool's own characters. Type your idea in a line, or pick a few characters and get five story ideas, or press 🎲 Surprise me for an unexpected mashup — a proud samosa against a sweet jalebi, live on the news. Choose the length, a mood (or describe your own, like “romantic comedy” or “festive”), the language and the audience. Besides Hinglish, Hindi and English, the 🌐 More languages list has 22 more, including Tamil, Telugu, Bengali, Marathi, Gujarati, Urdu, Spanish and Japanese. You also pick who narrates: the classic Sutradhar, a friendly Didi, the comic Masti, a dreamy Pari for fairy tales, Chhotu, a kid telling the story to his friends, a calm Rishi Muni or a Kathavachak for mythology, or one of 20 🗣️ Desi Boli storytellers who tell it with a warm regional touch: a man's and a woman's voice for each of Haryanvi, Bhojpuri, Rajasthani, Punjabi, Gujarati, Marathi, Bengali, Hyderabadi, Pahadi and Bundelkhandi. The whole script arrives with every voice already set. An AI editor then checks the script and fixes slips before it reaches you. A 1-minute story costs 20 credits. On the free plan the Director keeps it to about 12 lines, so you can write a story and voice it the same day. To check a script you wrote yourself, press 🔍 Check in the Story tab. Want more laughs or more chills? 😂 Aur mazedaar and 👻 Aur darawna rewrite the same story funnier or scarier with the same characters, voices and ending, and Undo brings the old one back.

Step 2: Give every character a voice

Every name in your script joins the cast on its own, matched to a fitting character voice. You can change any of them in one tap.

There are 227 ready-made characters in 19 groups, and each one has a free preview, so listen before you choose. A few that work well for stories:

• Sutradhar for the narrator, or one of its four siblings: Sutradhar Didi, Masti Sutradhar, Pari Sutradhar and Chhotu Sutradhar. For horror, Sannata is slow and eerie.
• Rishi Muni or Kathavachak Pandit Ji to narrate mythology and devotional kathas.
• The 🗣️ Desi Boli group for a story told with a warm regional touch, never a caricature: Haryanvi Tau and Taai, Bhojpuri Babuji and Amma, Rajasthani Hukum and Dadisa, Punjabi Paaji and Bebe, Gujarati Dada and Ba, Marathi Ajoba and Aaji, Bengali Dadu and Thakuma, Hyderabadi Mamu and Khala, Pahadi Nana and Nani, and Bundelkhandi Kaka and Kaki. Pick one as the Sutradhar in the Story Director and the narration itself gets a few everyday words from that boli.
• The Panchatantra group for fables: Kaalu Kauwa, Dhenchu Gadha, Bhookha Bhediya, Bagula Bhagat, Kekda Kaka, Gauri Gaay and more.
• Golu, Chintu or Pinky for children, and Principal Madam, PT Sir and Monitor Mona for school stories.
• Dadi Maa for a warm grandmother telling a bedtime story.
• Raja Sher, Raakshas or Andhkaar for kings, monsters and villains.
• 25 shayars, from a comic shayar and a tafribaaz tapori to a Sufi mystic and a mushaira ustad, for poetry and Shayari reels.

A story can have up to 10 people. Contrast is what keeps a story easy to follow: a child, a grandmother and a deep-voiced narrator are instantly clear, while three similar adult voices are not. If you're making a series, keep the same names in every episode and your characters will sound the same every time.

Step 3: Add sound effects and music

Story mode has a built-in library of 104 sound effects, 66 ambience loops and 20 music tracks, and using them costs nothing. Doors, thunder, footsteps, a ringing phone, a heartbeat, rain, a night full of crickets, a barking dog, a pressure cooker, a baraat, Diwali crackers, a royal court, bhakti and Sufi music. Put a Sound: line where the sound should happen. Put a Music: or Ambience: line just before the first line it should play under; it keeps playing until you change it. Lines spoken while a cave, temple, haveli, palace hall, courtroom or dungeon ambience plays get that room's echo on their own (you can switch it off under the volume sliders).

If you'd rather not place them yourself, press ✨ Add sounds. It reads the story and adds sound, music and ambience lines where they fit, without touching a word of your script, for 8 credits per 3,000 characters.

The mixing is done for you. Music dips under the voices while someone is speaking, sounds land between lines, and every scene gets its pause. Sliders let you turn music, ambience and sounds up or down, or switch all of them off.

Use sounds the way a film does: rarely, and where they mean something. One sound every few lines lands; a sound on every line becomes noise.

Step 4: Make the audio and fix single lines

Press Generate. Each line is spoken in its own character voice and the whole story is mixed into one track that you can download as MP3 or WAV.

Listen to it once from start to finish. If one line doesn't sound right, open ✏️ Fix a line under the player and press ↻ Redo on that line. Change a word or the [cue] if you like, or keep it as it is for a fresh take, and make it again. Only that line is remade, usually for a credit or two, and every other line stays exactly as it was. You can switch between the old and new take for free, and the fix works later too: open the story from your history and press ✏️.

Stories are made in Studio quality, which costs 30 credits per 1,000 characters. Each line is priced on its own and rounded up to a whole credit, so a three-minute story of about 2,400 characters in 60 lines costs about 100 credits. A story written in Hindi letters (Devanagari) costs a little more, because each Hindi letter says more than an English one and counts as 1.25 characters. The tool shows the cost before you start. On the free plan you can use up to 50 credits a day in this tool and a story can have up to 20 lines, about a minute of story. If a story needs more than what's left of your day, the tool tells you before it makes anything, so you're never charged for half a story. Paid plans don't have the daily or line limits.

Step 5: Plan the scenes

When your story is ready, open Make a video. Choose a look (3D animated, storybook painting, 2D cartoon, anime-style, realistic, claymation, comic book, pencil sketch, Madhubani, watercolour, or describe your own, such as Warli art) and a size: 16:9 for YouTube, 9:16 for Shorts and Reels, or 1:1.

Press ✨ Plan the scenes. The AI reads your story along with its exact timings and decides every shot: who is in it, what happens, and how long it stays on screen. It also writes one fixed description for every character and every place. Under Pictures, keep 🖼️ Every line and every single line gets its own picture prompt, so no moment of the story is left without a visual; choose Fewer for one picture every 7 to 12 seconds. A plan costs 18 credits per 10 lines with a picture for every line, or 8 per 10 lines with fewer pictures.

What you get is a prompt pack:

• a character sheet prompt with every character in one picture,
• a place sheet prompt for each location,
• a prompt for every scene, with how many seconds it lasts.

Step 6: Make the pictures and keep the faces the same

The most common problem in AI story videos is a hero who looks like a different person in every scene. The prompt pack is built to stop that, as long as you use it in order:

1. Make the character sheet first.
2. Then make each place sheet.
3. Then make the scenes in the same chat, attaching the character and place sheets each time.

Same chat plus same sheets gives you the same faces and the same rooms. Use the scene prompts as they are written. They describe what moves and who is where, and they never ask for text inside the picture, because your subtitles come from the tool.

Want a few scenes to move? Open 🎞️ Video prompts for image-to-video prompts (10 credits per 20 scenes). The scenes where motion adds the most are marked ⭐. These clips are silent on purpose; your story's own voices and sounds play over them.

Step 7: Upload, arrange and make the MP4

Upload up to 400 pictures and clips in any order, with any file names — enough for a picture on every line. Then either press ✨ Arrange with AI, which matches every picture to its scene (25 credits per 25 pictures, plus 1 per 20 scenes for each batch of up to 100 pictures), or use file-name order for free. Anything you want to move, you can drag.

Already have pictures or clips? If you made them somewhere else, you don't need a plan at all. In step 1 press 📁 Use my own pictures, drop them in any order and press ✨ Fit to the story with AI: it looks at each one, puts it on the line it shows, and leaves out anything that doesn't belong, like a character sheet. If your files are already named in order, Use in name order spreads them over the story for free.

Before you make the video, choose what goes on screen:

• Subtitles with each speaker's name in their own colour, and a title.
• Effects from the sounds, each one landing right on its sound: lightning when the thunder hits, a shake when a door slams, a tube light flickering on, confetti for applause, and a ✔ or ✖ for quiz answers.
• Looks from the ambience: rain, fog, fireflies on a jungle night, petals at a wedding, gulal at Holi, fireworks at Diwali, an underwater glow, stars in space. A night look goes easy on pictures that are already dark.
• Mood colours from the music: cooler for scary, warmer for happy, golden for mythology, purple for magic.
• 🎭 Camera follows the acting: a [whispering] line moves in close, [shouting] jolts, [laughs] bounces and [sad] fades the colour.
• 💥 Comic words & emoji, off until you tick them: BOOM! and DHISHOOM! burst out of a comic-book explosion, BHOW BHOW! pops in word by word, and a laugh gets HA HA! with a 😂 sticker. In a story written in Devanagari they come out as धमाका!, ढिशूम! and भौं भौं!.
• Strength: Gentle, Normal or Strong. Open ✨ Effect moments to see every effect with its time, and press ✕ on any one you don't want.
• 🛡️ Flash-safe: lightning, bangs and flickers never flash more than 3 times a second, even at 1.5× speed, and the dips to black between scenes count towards that limit too. When sounds come too close together, the extra ones glow or dim softly instead.
• Fades at the start and end, and a dip to black between scenes, each with its own switch.
• Speed from 0.9× to 1.5×.
• Quality: Full HD 1080p on every plan, or ✨ 4K with Pro on a laptop or computer, for videos up to 15 minutes at the speed you pick (10 minutes on a computer with under 8 GB of memory). YouTube plays 4K uploads in sharper quality, even on phones.
• Your channel: add your logo once and it sits in a corner of every video, and tick 🔔 Subscribe card at the end for a 4-second card with your logo and channel name.

Press Make the video. The video is put together on your own device, so this step costs no credits, and you download an MP4. Here are 40 seconds of a story made this way:

Under the finished video, the ▶️ YouTube pack has what the upload needs: ⬇ Subtitles (.srt) to add as closed captions, 📑 Copy chapters for the description (both free), ✨ Title, description & tags written by AI from your story for 10 credits per 5,000 characters of the story, and a shortcut to the AI Thumbnail Maker. If you download the video at another speed, the subtitles file is timed to match it.

Stories in Tamil, Bengali, Marathi and 19 more languages

Nothing in this guide is only for Hindi. The voices speak 25 languages: Hindi, Hinglish and English, plus Marathi, Bengali, Tamil, Telugu, Gujarati, Kannada, Malayalam, Punjabi, Urdu, Odia and Nepali, and Spanish, French, German, Portuguese, Italian, Arabic, Indonesian, Japanese, Korean, Russian and Turkish. Write the story in the language you want, or pick it in the 🎬 Story Director and let it write the story for you.

Everything else stays the same. Speaker names, [cues] and Sound:, Music: and Ambience: lines are still written in English letters; only the spoken words change. 🔍 Check works out the language from the script itself before it looks for slips, and the subtitles in the video use the story's own script.

What a three-minute story video costs

A three-minute story is usually about 60 short lines with some 2,400 spoken characters. With a picture for every line, it costs:

Without the optional steps it comes to about 210 credits. With Fewer pictures, about 20 for three minutes, the plan costs 48 and arranging 26, so the total is about 215, or about 150 without the optional steps. A story in Hindi letters (Devanagari) adds up to a quarter to the audio.

Other AI helpers cost extra only when you use them: five 🎬 story ideas 13 credits, ✨ Add sounds 8 per 3,000 characters, 🔍 Check 6 plus 2 per 1,000 characters, and a 😂 funnier or 👻 scarier rewrite 19 per 3,000 characters. The pictures cost whatever your image tool charges.

Seven mistakes that make story videos look cheap

1. Long lines. Split them. One breath, one feeling.
2. Too many people in one scene. Two or three speakers per scene is easy to follow.
3. Loud music. Pick a calm track for scenes with a lot of talking.
4. Changing faces. Always make the character sheet first, and reuse it.
5. Text inside pictures. Let the subtitles do that job.
6. A sound on every line. Save sounds for moments that matter.
7. No scene breaks. Blank lines give listeners a moment to breathe before the next scene.

Make the next episode faster

Keep the same character names and they keep the same voices. Keep your place sheets and reuse them for places that come back. Your last 20 results stay in your history, and opening the video studio again from there brings back your scene plan without paying for it twice.

Start your first story

Write a short story of five or six lines first, just to hear your cast. Then make the full one. Open Story mode and start with the free plan's 200 credits, or read more about the AI Voice Generator and listen to the other samples there.