Diwali shop offer in English and Hindi: one clip, two voices, $2.13
One 12 s Wan 3.0 clip, an English and a Hindi voice-over, two Timeline renders and two caption jobs for a Diwali offer: $2.1266 in total on Sume.
A Diwali offer for a shop with English and Hindi speaking customers can share one visual and carry two voice-overs. On Sume that is one 12-second Wan 3.0 clip at $1.50, two text-to-speech jobs of about 280 characters at $0.0133 each, two Timeline renders at $0.10 each and two caption jobs at $0.20 each: $2.1266 for both language versions.
The existing Diwali posts cover a photo greeting and a five-day series. This one is a single offer clip in two languages, for the case where the visual stays and only the voice and text change.
One clip for both versions
Start from a still of the shop window with the diyas lit. Animate it for 12 seconds with Wan 3.0 at 720p: $0.125 per second, $1.50 for the clip. Because Video models do not lip-sync to generated speech, keep people out of the frame or turned away; the Models overview states that speaking shots belong to the avatar route (models overview). A lit window, sweets and a counter need no lip-sync.
Two voice-overs
Sume's TTS 1.0 is $0.0475 per 1,000 characters, with spaces and punctuation counted and a limit of 20,000 characters per job. A 280-character script is $0.0133. For the Hindi version, write the script in Devanagari, choose a Hindi voice from the voice library and set language to hi. The TTS schema notes that the provider assumes English when language is omitted, so set it for every non-English script. generation_config.speed accepts 0.6 to 1.5, so you can fit a line to the clip; check the audio length against 12 seconds before you render.
The Hindi voice list and language codes are covered in a separate post on Chinese, Hindi and Russian voices.
Join and caption
Each version is one Timeline render with the voice as the audio spine and the clip as the single video slot, 12 seconds long, so $0.10 each (Timeline). Caption each with script_text set to the exact line, which aligns your words to the speech instead of trusting recognition: $0.20 each (video captions). Test one Hindi caption on a short clip first, because the Devanagari font coverage is covered in its own post and you should see the result before burning 100 of them.
| Line | Basis | USD |
|---|---|---|
| Wan 3.0 clip, 12 s, 720p | $0.125 per second | $1.50 |
| TTS, English, 280 chars | $0.0475 per 1,000 | $0.0133 |
| TTS, Hindi, 280 chars | $0.0475 per 1,000 | $0.0133 |
| 2 x Timeline render | $0.10 each | $0.20 |
| 2 x caption job | $0.20 each | $0.40 |
| Total | $2.1266 |
Adding a third language
Gujarati, Tamil or Marathi would each add one TTS job, one render and one caption job: about $0.31 per language on top of the shared clip, using the same 280-character script length. Check that the voice library has a voice for the language before you plan it.
Related posts
More in Use cases
- DOOH 1080x1920 10 s master: a render checklist
An example vertical DOOH master (1080x1920, H.264 MP4, 10 s, 25 to 30 fps). What a Sume video request controls and what you must check yourself.
- eBay AI listing photos: the seller owns accuracy
eBay's User Agreement makes sellers responsible for listing accuracy, including generative AI content. Generate 1:1 stills from your real photo and check them.
- E-commerce product video cost per SKU: cutout plus a 5-second clip
A cutout at $0.0225 plus a 5-second 720p Seedance 2.5 clip is $2.9115 per SKU on Sume. For 100 SKUs that is $291.15. Arithmetic inside.
- Etsy 'Designed by' AI disclosure: keep a job record per listing
Etsy asks sellers to mark AI-assisted items as designed by the seller and disclose it. Keep the prompt and Sume job id for each listing as your record.
Written by Sume