Safety module in 10 languages: 60 seconds, 10 caption jobs, $2.00

A 60-second safety module captioned in ten languages costs about $2.02 on Sume: ten caption jobs at $0.20, one audio detach at $0.01 and one speech-to-text minute at $0.01 to get timed words once. The translation runs on Bilibili's listed free Index-Translate API, which covers 150 languages.

This assumes one 60-second spoken English module and ten target languages that you have tested for caption glyphs. Prices are from the Video captions docs, audio detach docs and pricing page, read 2026-10-05.

Where the $2.02 goes

The transcript is paid once and reused for all ten languages.

Turn STT words into cues once

The speech-to-text result carries words[] with word, start and end in seconds. Group them into short cues, translate each group, and burn the translated cues, because script_text aligns your script to the speech timings and is meant for the spoken language.

words = [{"word": w, "start": i * 0.4, "end": i * 0.4 + 0.35}
         for i, w in enumerate("Always wear gloves when you handle sharp tools near the line".split())]

def to_cues(words, size=6):
    cues = []
    for i in range(0, len(words), size):
        chunk = words[i:i + size]
        cues.append({"text": " ".join(w["word"] for w in chunk),
                     "start": chunk[0]["start"], "end": chunk[-1]["end"]})
    return cues

for cue in to_cues(words):
    print(cue)

Compliance still needs a person

A machine translation of a safety instruction is a draft. Have a fluent colleague check each language before the module reaches a shift, and keep the English version on the same page.