Google Cloud TTS caps requests at 5,000 bytes; Sume counts characters

Google Cloud Text-to-Speech lists "Total bytes per request: 5,000" and notes that multi-byte characters, such as ja-JP, count toward the limit. A Japanese or Korean script therefore fits fewer characters than an English one. Sume TTS 1.0 counts characters, not bytes: 1 to 20,000 per request, with spaces and punctuation included.

Google's limits are from its quotas page, read 2026-10-01; Sume's from the OpenAPI schema behind the API reference.

Why does bytes versus characters matter?

Because one character can take more than one byte. In UTF-8, a Latin letter is one byte while a Hangul syllable or a kana is three. So 5,000 bytes holds 5,000 English letters but roughly 1,666 Hangul syllables. Count bytes, not characters, if you target a bytes-based limit.

How do the limits compare?

Two units, same job.

What do I send for Korean on Sume?

Set language. The schema says to set it for every non-English transcript; Sume infers Korean or Japanese from a transcript that is only Hangul or kana, but an explicit value is safer.

curl -X POST https://api.sume.com/v1/tts-1.0/generate \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "transcript": "안녕하세요. 오늘의 소식을 전해드립니다.",
    "avatar_handle": "@narrator",
    "language": "ko"
  }'

Do I still have to split long text?

Only past 20,000 characters or 1,200 seconds. See text to speech for long text for chaptering and joining.