talk_show_clip.mp4 · 58 s
اما she is frustrated على خاطر he is shorter than her
Tapra-Voice
Batch Tunisian speech to text, with the code-switching intact.
POST /v1/audio/transcriptions
$0.050 per audio minute
The model family
Speech in, script normalized, meaning moved between languages, and a Tunisian voice coming back out. Each model has its own endpoint and its own metered price, and every example on these pages is real, unedited output.
Pick a model
The line on every card is a recorded request: the input sent, and the output the model returned.
talk_show_clip.mp4 · 58 s
اما she is frustrated على خاطر he is shorter than her
Batch Tunisian speech to text, with the code-switching intact.
POST /v1/audio/transcriptions
$0.050 per audio minute
3lech 7abit l9anoun?
علاش حبيت القانون؟
Arabizi ⇄ Arabic script, without eating your numbers.
POST /v1/converters/transliterate
$0.033 per 1,000 characters
عندي برشا خدمة اليوم، نكلمك بعد ال réunion.
J’ai beaucoup de travail aujourd’hui, je vous appellerai après la réunion.
Derja in, real meaning out — in both directions.
POST /v1/translations
$0.050 per 1,000 characters
أهلا خويا، نتلاقاو غدوة après le travail؟
instruct: "Male, Young Adult, Moderate Pitch"
A Tunisian voice that sounds like it grew up here.
POST /v1/audio/speech
$0.067 per 1,000 characters
open → 100 ms PCM frames → {"type":"stop"}
ready 0.01 s → first partial 1.97 s → utterance 23.64 s → final 28.06 s (reason: client_stop)
Partial tokens while the sentence is still being spoken.
WSS /v1/realtime/stream
$0.100 per audio minute
All five models draw from the same prepaid balance, metered per audio minute or per 1,000 characters.
Start with 2 TND freeChain them
Transcribe the call, move the notes into French, and answer with a Tunisian voice.
Stream partial transcripts over WebSocket and translate each turn once it locks.
Normalize Arabizi into Arabic script, then translate it for a wider audience.