Languages

clipsdk was tested on talks in 14 languages, from Hindi and Arabic to Japanese, with captions in the right script. Every language bills the same $0.03 per source minute, and $1 of free credit covers about 33 minutes.

AI clipping in Hindi, Spanish, Portuguese and 11 more languages

Each of the 14 rows below is a real clip from a freely licensed talk, with its caption preset, checked frame by frame.

Get API key Pricing

Updated 7 November 2026

Or one API call

$0.03per source minute

Every clip included, up to 1080p60 and 20 Mb/s
$1 free to start, 33 minutes of video
Prepaid credit, no subscription

See pricing

14 languages, one real clip each

Tap a clip to play it with sound. Try it opens the Hindi, Spanish or Portuguese talk in New clips with the language set.

Hindi
हिन्दी
bold, 36 s
Try it
Spanish
Español
karaoke, 17 s
Try it
Portuguese
Português
bold, 34 s
Try it
Indonesian
Bahasa Indonesia
karaoke, 16 s
French
Français
bold, 23 s
German
Deutsch
karaoke, 40 s
Japanese
日本語
bold, 39 s
Chinese
中文
bold, 29 s
Korean
한국어
bold, 19 s
Arabic
العربية
bold, 36 s
Hebrew
עברית
karaoke, 26 s
Urdu
اردو
karaoke, 28 s
Russian
Русский
bold, 27 s
Turkish
Türkçe
karaoke, 39 s

Tested languages and their clips

Each clip link opens the MP4 the job rendered.

Tested languages and their clips
Language	Script	Clip	Captions	What came out right
Hindi	Devanagari	36 s MP4	`bold`	Conjuncts joined (अच्छा), Hindi transcript
Spanish	Latin	17 s MP4	`karaoke`	Accents kept (inglés, papá), word by word sweep
Portuguese	Latin	34 s MP4	`bold`	Accents kept in capitals (PORTUGUÊS)
Indonesian	Latin	16 s MP4	`karaoke`	Clean Indonesian transcript, word by word sweep
French	Latin	23 s MP4	`bold`	Accented capitals kept (SÉANCES)
German	Latin	40 s MP4	`karaoke`	Umlauts kept (München, einschätzen)
Japanese	Kanji and kana	39 s MP4	`bold`	Kanji and kana, no spaces added
Chinese	Simplified Han	29 s MP4	`bold`	No spaces, F1 kept in Latin letters
Korean	Hangul	19 s MP4	`bold`	Hangul with word spacing kept
Arabic	Arabic	36 s MP4	`bold`	Letters joined, words right to left
Hebrew	Hebrew	26 s MP4	`karaoke`	Right to left, words in reading order
Urdu	Arabic	28 s MP4	`karaoke`	Urdu letters (ے ں ہ) joined, right to left
Russian	Cyrillic	27 s MP4	`bold`	Cyrillic in capitals (ДРУЖИТЬ С СЕМЬЯМИ)
Turkish	Latin	39 s MP4	`karaoke`	Turkish letters kept (Şeffaf, örgütlerinin)

How we test a language

Every row passed the same 3 checks. A language that fails one stays off this page.

Pick a freely licensed talk
Each source is a talk or interview from Wikimedia Commons under CC BY, CC BY-SA or public domain, mostly one speaker. We upload the file or send its direct URL, like you would.
Run one job on the live API with the language set
Each job asks for 3 clips of 15 to 40 s, with language set and the bold or karaoke caption preset.
Check the frames by eye
We pull frames from each clip with ffmpeg and look for broken letters, empty boxes and cut-off lines, then read the transcript. Right-to-left words must stay in reading order.
frame.sh
```
curl -sO https://api.clipsdk.com/f/out/job_n351mgngyrlm2ik21ca7/clip-1.mp4
ffmpeg -ss 20.2 -i clip-1.mp4 -frames:v 1 hindi-frame.png
```

Set the language in one request

Send language as an ISO 639-1 code like hi, es or pt. Leave it out and the spoken language is detected. A region code is fine too: pt-BR reaches Whisper as pt.

This is the exact request behind the Hindi row: 3 clips from a talk just under 2 minutes long, with bold captions.

POST/v1/clipsTry it

curl https://api.clipsdk.com/v1/clips \
  -H "Authorization: Bearer $CLIPSDK_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "video_url": "https://api.clipsdk.com/f/lib/demo/wikitongues-krishna-hindi.mp4",
    "clips": 3,
    "min_seconds": 15,
    "max_seconds": 40,
    "captions": "bold",
    "language": "hi"
  }'

How the captions handle each script

Whisper large-v3-turbo writes the transcript. The bold preset sets Montserrat in capitals, 3 words at a time. The karaoke preset uses Poppins, 5 words at a time, and sweeps a highlight across each word.

Japanese, Chinese, Korean, Arabic, Urdu, Hebrew and Hindi switch to a matching Noto Sans face. Arabic, Urdu and Hebrew run right to left, and Japanese and Chinese lines join without spaces. Latin stays in the preset font, and so does Cyrillic with bold. Accents came through in every Latin-script language we tested.

Hindi clips for clipping campaigns

Whop describes Content Rewards as view-based campaigns that pay creators only for actual views generated, either by clipping a brand's existing videos or by making new ones.

For a clipping team in India, clipsdk does the cutting by API: send a long Hindi video, get ranked 9:16 clips back with Devanagari captions. The price per source minute is the same as English, and new accounts start with $1 of credit.

Questions, answered

Does clipsdk make clips in Hindi?

Yes. A Hindi talk just under 2 minutes long came back as 3 ranked clips with Devanagari captions and joined conjuncts. Set language to hi, or leave it out and the language is detected.

Which languages can I use?

The 14 in the table each passed a real render with its captions checked by eye. A language that is not in the table is untested, so try it on a short file before you build on it.

Is clipsdk an OpusClip alternative for Hindi?

clipsdk is an API, not an editor. You send a long Hindi video by upload or direct file URL and get ranked vertical clips with Hindi captions back, for $0.03 per source minute.

What is Whop clipping (whop clipping kya hai)?

Whop's Content Rewards lets brands pay creators for actual views. In a clipping campaign, clippers cut a brand's long videos into short posts. clipsdk does that cutting, reframing and captioning by API.

Does a Hindi or Arabic clip cost more than English?

No. Every language bills $0.03 per source minute, and new accounts get $1 of credit, about 33 source minutes.

Sources

Krishna speaking Hindi, Wikitongues and Teddy Nee, CC BY-SA 4.0, checked 7 November 2026
Regina speaking Spanish, Wikitongues, CC BY 3.0, checked 7 November 2026
Freddie speaking Portuguese, Wikitongues, CC BY 3.0, checked 7 November 2026
Nila speaking Indonesian, Wikitongues, CC BY 3.0, checked 7 November 2026
Interview with Sarah Hanffou, Cornilleau, CC BY 3.0, checked 7 November 2026
Gereon speaking German, Bogreudell, CC BY-SA 4.0, checked 7 November 2026
Fumio Kishida's press conference, Cabinet Public Relations Office of Japan, CC BY 4.0, checked 7 November 2026
Zhou Guanyu on F1 and China, China News Service, CC BY 3.0, checked 7 November 2026
Hanbid speaking Korean, Wikitongues and Teddy Nee, CC BY-SA 4.0, checked 7 November 2026
First address of the Prime Minister, Prime Minister's Media Office, CC BY 4.0, checked 7 November 2026
Tel Aviv University experts answer questions on the coronavirus, Tel Aviv University, CC BY 3.0, checked 7 November 2026
Memories of Ramadan with Khalid Masood Khan, VOA Urdu, public domain, checked 7 November 2026
Alexander Rosenbaum interview, VOA, public domain, checked 7 November 2026
Istanbul Medical Chamber press statement, VOA, public domain, checked 7 November 2026
Whop: Content Rewards, checked 7 November 2026