Turn Arabic text into speech
Loading
Fast replies, natural conversation
Add real-time Arabic speech to your agent or assistant for fast replies and conversations that flow.
LiveKit and API integrationsA dialect familiar to your customers
Choose an MSA or regional Arabic voice that suits your audience, for agent replies or content recordings.
Try your own text
- 01
Enter a short Arabic script
Replace the demo text with your message. Add punctuation for pauses and spell out amounts when their reading matters.
- 02
Choose a voice and listen
Choose a voice and press Listen.
Check the words your listeners need to hear
These fictional script prompts help you compare voices. Keep the script the same while comparing, and choose MSA or regional wording separately from the voice.
Names and mixed-language terms
Try: «مرحبًا سارة، وصل طلبك من Sawtak Arabi». Replace the name and brand with the ones your listeners will hear. If a name sounds wrong, try a clear Arabic spelling and compare.
An amount and a next step
Try this formal script: «قيمة الطلب مئة وخمسون ريالًا سعوديًا. راجع التفاصيل ثم أكّد طلبك». Check the currency and the call to action. Use dirhams for a UAE message when appropriate.
Punctuation and local wording
Break long sentences at natural pauses. Use the wording your audience expects, then compare a suitable regional voice with MSA if both fit the project.
Listen through the ending
Check the whole take in the context where it will be used. If one sentence is unclear, revise that sentence and generate again; keep the accepted recordings for the rest of the project.
Arabic text to speech API: your first request
Create an API key, set SAWTAK_API_KEY on your server, and replace <VOICE_ID> with a ready ID from GET /v1/voices. This request saves a WAV stream and response headers. Follow the quickstart to verify completion and finalize the file before using it.
curl --fail --show-error https://api.sawtakarabi.ai/v1/audio/speech \
-H "Authorization: Bearer $SAWTAK_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "arabic-tts-1",
"voice": "<VOICE_ID>",
"input": "مرحباً بك",
"response_format": "wav"
}' \
--dump-header speech.headers \
--output speech.partial.wavWhat to check before integrating
Start with short requests. The API accepts up to 10,000 characters per request and streams PCM or WAV. Speech is charged for the full input once audio is delivered, including interrupted streams. Check current rates and retry guidance before running a larger workload.
Make and revise a video with Remotion
For coding-agent users: bring your narration WAV files to an editable Remotion project. The guide covers rendering and replacing one scene’s audio.