Sign text
Send text, get the signs back a word at a time. The first word arrives while the rest of the sentence is still being signed.
The request
/v1/sign| Field | Required | Meaning |
|---|---|---|
text | Yes | What to sign. Punctuation is dropped; each word is signed in order. |
from | No | The language of the text: rw, en, fr, sw or ja. Default rw. |
to | No | The sign language. Default ksl (Rwandan Sign Language). All languages. |
curl https://api.usesigna.org/v1/sign \
-H "Authorization: Bearer $SIGNA_KEY" \
-H "Content-Type: application/json" \
-d '{"text": "Good morning, welcome to the news", "from": "en", "to": "ksl"}' \
--output news.sgnThe response
200 with Content-Type: application/vnd.signa.sgn. The body is one .sgn segment per word, written the moment that word (and every word before it) is ready. Read the body as a stream to start playing the first word straight away; read it to the end and you have the whole sentence as one valid .sgn file.
A word that cannot be signed is skipped, so one unusual word never costs you the sentence. Only when no word at all could be signed do you get an error instead.
Speed
Every word is looked up at once. Words signed before, by anyone, come back almost instantly; new words are prepared in parallel. A sentence of everyday words usually starts playing in well under a second.
Long texts
Billing
The characters of text count against your monthly character limit, once, after something was signed. A request that fails costs nothing.
The whole sentence as .pose
/v1/spoken-to-signed-pose?text=…&spoken=en&signed=kslThe same signs as one standard .pose file, for tools that already read that format. It answers once the whole sentence is ready, so /v1/sign is the better choice for anything shown live.