Speech to text

Transcribe a recording, then sign the text. For speech as it happens, use live signing instead.

Transcribe a recording

POST/v1/stt

A multipart form with the recording in audio (MP3, WAV, M4A, WebM or OGG, up to 25 MB) and its language in lang.

POST /v1/stt

curl https://api.usesigna.org/v1/stt \
  -H "Authorization: Bearer $SIGNA_KEY" \
  -F audio=@ikiganiro.m4a \
  -F lang=rw

Response

{
  "success": true,
  "payload": { "text": "Muraho neza", "lang": "rw", "provider": "local" }
}

Send text to /v1/sign to sign it. Transcription itself is not counted against your characters; the signing is.

Language	lang	Speed
Kinyarwanda	`rw`	About half a second
Swahili	`sw`	About half a second
English	`en`	One to two seconds
French	`fr`	One to two seconds
Japanese	`ja`	One to two seconds

Speak text aloud

POST/v1/tts

The other direction, for apps that also speak: send {"text", "lang", "speechRate"} and get {"audio", "provider", "lang"} back, with the audio as base64.

← Subtitles Live signing →