Korean to Japanese voiceover: translate, then TTS with language ja

Sume TTS speaks the text you send and never translates it, so a Korean-to-Japanese dub is three steps: transcribe the Korean audio, translate the text yourself, then send Japanese text with language set to ja. A 400-character Japanese line costs $0.01900 at $0.0475 per 1,000 characters.

Pipeline and per-step cost

Example: a 2-minute Korean clip and a 400-character Japanese script. STT uses the documented $0.01 per audio minute.

The TTS step

Pick an avatar whose voice speaks Japanese. If the voice and language disagree, Sume returns the mismatch warning described in confirm_language_mismatch; do not set that flag blindly.

{
  "transcript": "今日のおすすめは、軽くて丈夫なレインジャケットです。",
  "avatar_handle": "@narrator-ja",
  "language": "ja",
  "generation_config": {"speed": 1.0},
  "metadata": {"source_lang": "ko", "target_lang": "ja"}
}

Gotchas

Japanese is dense per character, so the same meaning is often fewer characters than the Korean or English version; price by the final string. Timing will differ from the source clip, so use word timestamps to retime, or tune generation_config.speed. The STT line is arithmetic on the documented per-minute rate; check usage on your own job.