# How to add Hinglish captions to videos

Canonical URL: https://getvidiyo.app/blog/hinglish-captions
Last updated: 2026-10-08
Author: Ajay Pawriya (https://getvidiyo.app/about)
Published: 2026-10-08
Category: Captions

> TL;DR: Transcribe with a model that understands Hindi, then convert the Hindi words to Roman script and fix the spelling by hand. Speech models write Hindi in Devanagari, so no free tool gives you clean Roman Hinglish straight away. Vidiyo transcribes Hindi locally and keeps your word fixes through every cut.

I talk in Hinglish on camera, like a lot of creators in India. Captioning those videos is where most tools quietly fall apart: the speech model hears the Hindi fine, then writes it in Devanagari, or mangles it into English words that sound close. This guide is the workflow I use, with the parts that still need a human called out plainly.

## What are Hinglish captions?

Hinglish captions are captions for speech that mixes Hindi and English, usually written in Roman script. [Hinglish](https://en.wikipedia.org/wiki/Hinglish) switches languages freely inside a sentence ("yeh feature launch ho gaya, finally"), and when people type it, they mostly use the English alphabet. Captions in that style read the way your viewers already text.

The alternative is captioning the Hindi parts in Devanagari, which is correct for Hindi but looks unfamiliar to a lot of Hinglish audiences next to English words.

## Why is Hinglish hard for speech models?

Speech models pick one script per language, and Hinglish doesn't fit that. A Whisper model set to Hindi writes Hindi words in Devanagari. An English-only model isn't built for Hindi, so Hindi words come out as wrong English guesses. Neither gives you clean Roman Hinglish on the first pass.

So any honest Hinglish workflow has two steps: transcribe accurately, then convert and correct the script. The goal is to make the second step as short as possible.

## Which caption script should I use?

Choose the script your audience reads fastest. For most Hinglish creators that's Roman script; for Hindi-first audiences it's Devanagari. Here's how the options compare:

| Option | What viewers see | Effort after transcription | Good for |
|---|---|---|---|
| Devanagari for Hindi words | "यह feature launch हो गया" | Low: check words | Hindi-first audiences |
| Roman Hinglish | "yeh feature launch ho gaya" | Medium: convert and fix spellings | Most Hinglish creators |
| English only | "this feature is live now" | High: needs translating | Not a caption, a translation |

## How do I add Hinglish captions in Vidiyo?

Pick the spoken language, download the language pack once, generate captions, then convert the Hindi words to Roman script and fix them in the Transcript tab. Transcription runs on your Mac, and every fix stays attached to its word, so later cuts don't undo your work.

1. **Set the language.** Open **Settings → Transcription** and choose **Hindi**, or **Detect automatically**, which the app suggests for Hinglish. If your speech is mostly English with a few Hindi words, try English and compare.
2. **Download the language pack.** Non-English languages use a multilingual model of about 575 MB. It downloads once; after that it works offline. See [the language pack docs](https://getvidiyo.app/docs/language-pack).
3. **Generate captions.** In the Captions tab, click **Generate captions**. Hindi words come out in Devanagari.
4. **Convert and fix.** Edit words in the **Transcript** tab. If you've connected Claude or ChatGPT, you can ask: "Rewrite the Hindi caption words in Roman script, Hinglish style, and keep the English words as they are." It changes the text of each word without touching its timing. Then read every line yourself, because spelling choices are yours and the assistant can get some wrong.
5. **Style.** Short Hinglish clips usually suit Word highlight or Pill karaoke at three words a group. See [word-by-word captions](https://getvidiyo.app/blog/word-by-word-captions).

Two things make this hold up over a real edit. Corrections are stored per word, not per caption block, so changing the style or regrouping keeps them. And because captions are anchored to the spoken words, cutting a sentence later just removes its captions instead of shifting everything after it.

## How do I add Hinglish captions without Vidiyo?

Use platform auto captions or Whisper on the command line, then fix the script in a text file. Both work and both cost nothing, but you'll do the conversion by hand.

1. **Transcribe.** [YouTube](https://support.google.com/youtube/answer/6373554) lists Hindi among its automatic caption languages, and [Instagram](https://techcrunch.com/2022/03/01/instagram-is-rolling-out-auto-generated-captions-for-videos/) launched auto captions in 17 languages including Hindi. Offline, run whisper.cpp with `-l hi -osrt` to get an SRT in Devanagari.
2. **Convert.** Open the SRT in a plain text editor and rewrite the Hindi lines in Roman script. Leave the numbered lines and timestamps untouched.
3. **Deliver.** Upload the SRT to YouTube as a caption track, or burn it into the video with your editor or FFmpeg.

Platform auto captions have a second limit: you can't style them, and they only exist on that platform. For Reels with branded, word-by-word captions, you'll need them in the video itself.

## How should I spell Hinglish consistently?

Pick one spelling for each common word and stick to it across every video. There's no official Roman spelling for Hindi, so consistency is what makes captions look deliberate. Keep a short style sheet and paste it into your notes, or give it to the assistant along with the request.

| Word | Pick one | Avoid mixing with |
|---|---|---|
| है | hai | hain, he |
| नहीं | nahi | nahin, nai |
| क्या | kya | kyaa |
| अच्छा | accha | acha, achha |
| भाई | bhai | bhaai |
| यार | yaar | yar |
| ठीक है | theek hai | thik hai |
| मतलब | matlab | mtlb |

A few more rules that save time:

- **Keep English words in English.** "Feature", "launch", "deadline" stay as they are. Don't phoneticise them.
- **Use the spelling your audience comments in.** Scroll your comments and copy their habits.
- **Don't add diacritics.** Roman Hinglish on social media doesn't use them, and they make captions harder to read at speed.
- **Fix names first.** People's names, product names and places are the most visible mistakes when they're wrong.

## How long does the fix-up pass take?

It depends on how much of the video is Hindi and how clearly you speak, so I won't give you a number. What helps most is cutting before you caption: remove silences, filler and retakes first, so you only correct words that make the final video. A tight 45-second Reel has far fewer words to check than the raw take it came from.

If you post the same video in English and Hinglish, caption each cut separately rather than translating one into the other. Vidiyo doesn't translate, and a translated caption track wouldn't match your Hinglish speech anyway.

## What's the honest state of Hinglish captions?

They're good enough to be worth doing and not yet automatic. Transcription of Hindi speech is solid; the script is the gap. Until speech models write Roman Hinglish natively, budget a review pass for every video, and keep a spelling sheet so that pass gets shorter over time. More on making videos for Indian audiences on the [Indian creators page](https://getvidiyo.app/for/indian-creators).

## Honest limits

- Hindi transcribes in Devanagari, so Hinglish captions need some words fixed by hand.
- Hindi and other non-English languages need the language pack, a one-time download of about 575 MB.
- Vidiyo doesn't translate captions.

## Frequently asked questions

### Can AI transcribe Hinglish directly into Roman script?

Not reliably. Whisper-based models write Hindi words in Devanagari, and English-only models aren't built for Hindi at all. Expect to convert and fix some words by hand.

### Which language should I pick in Vidiyo for Hinglish?

Vidiyo's Settings suggest trying Detect automatically for Hinglish. If your video is mostly Hindi, choose Hindi; if it's mostly English with a few Hindi words, English may need fewer fixes.

### Should Hinglish captions be in Devanagari or Roman script?

Match how your audience types. Most Hinglish creators caption in Roman script because that's how their viewers write comments and DMs, but Devanagari is right for Hindi-first audiences.

### Does Vidiyo translate Hindi captions to English?

No. Vidiyo captions what was said, in the language it was said. It isn't a translation tool.

### Do YouTube and Instagram auto captions support Hindi?

Yes, both list Hindi among their auto-caption languages. They have the same script issue, though, and you can't style their captions.

## Related

- [Vidiyo for Indian creators: Hindi, Hinglish and Indian-language videos](https://getvidiyo.app/for/indian-creators)
- [How do I download the language pack?](https://getvidiyo.app/docs/language-pack)
- [Auto captions that are transcribed on your Mac](https://getvidiyo.app/features/auto-captions)
- [How to make word-by-word captions](https://getvidiyo.app/blog/word-by-word-captions)
- [The best caption styles for Reels and Shorts](https://getvidiyo.app/blog/caption-styles-for-reels-and-shorts)
- [Whisper transcription](https://getvidiyo.app/glossary/whisper-transcription)

## Sources

- [Wikipedia: Hinglish](https://en.wikipedia.org/wiki/Hinglish) (accessed 2026-10-08)
- [YouTube Help: Use automatic captioning](https://support.google.com/youtube/answer/6373554) (accessed 2026-10-08)
- [TechCrunch: Instagram is rolling out auto-generated captions for videos](https://techcrunch.com/2022/03/01/instagram-is-rolling-out-auto-generated-captions-for-videos/) (accessed 2026-10-08)
- [whisper.cpp command-line options](https://github.com/ggml-org/whisper.cpp/blob/master/examples/cli/README.md) (accessed 2026-10-08)

Download Vidiyo for Mac (free during the beta): https://getvidiyo.app/download