Japanese text-to-speech API: voices, example and limits

Generate Japanese speech from text over plain HTTP: 2 live voices, no setup beyond an API key. Japanese text with kana and kanji worked in our test. There is a female and a male voice.

Status: every non-English result is marked native_review_pending until a native speaker approves the language. Treat the audio as unreviewed for anything customer-facing and have a native speaker listen before you ship it.

Voices

Name	Id	Gender	Description	Library clips
Alpha	`jf_alpha`	female	Japanese	19
Kumo	`jm_kumo`	male	Japanese	20

Hear the library

The library holds 23 checked Japanese clips for the live voices (short words and phrases). Press to play:

Speak a sentence

Anything outside the library is generated live. Our test sentence (Japanese, voice jf_alpha) took 10.0 seconds on 2026-10-08 through the public demo endpoint (one observation, no GPU worker on this path; do not treat it as a benchmark).

cURL

curl -X POST https://speakvora.com/api/v1/speak \
  -H "x-api-key: $SPEAKVORA_KEY" -H "content-type: application/json" \
  -d '{"text":"ご注文の商品は受け取りの準備ができています。","voice":"jf_alpha","lang":"ja"}'

# 200 {"url":"...","lang":"ja","voice":"jf_alpha","format":"mp3","live":true,"review_status":"native_review_pending"}

Write numbers as words

Outside English, digits are rejected. We tested this: ご注文番号12番の準備ができています。 returns invalid_text. Write the number out instead: ご注文番号十二番の準備ができています。. Remove placeholders such as (name) too.

Limits and honest notes

1 to 500 characters per request; the language you send must match the voice.
Live generation for this language runs on the standard lane in our tests; library clips and repeats are quick.
Your own voice is not offered in Japanese except where noted above; use a catalog voice.
Billing is per character at $0.70 per million for catalog voices, same as English. A non-Latin character counts as a character.

Is Japanese speech reviewed by a native speaker?

Not yet. Results carry review_status native_review_pending. English is our reference language.

Can I use my own voice in Japanese?

Custom voices support English, with French and German in beta. Spanish, Hindi, Chinese and Japanese were not intelligible in testing and are not offered.

How do I pick the language?

Send lang as an ISO code (for example ja) and a voice from that language. Mixing a voice and a different language is rejected.

Next: all languages · how generation works · billing · errors and retries

Japanese text to speech API