Iraqi, and more), native diacritics handling, right-to-left text, and Arabic-English code-switching in the same sentence.

Two models, one API:

Hakim Fast v1.3 — Streaming TTS with 113ms time-to-first-byte, 15 Arabic dialects, and 30+ languages including English, French, Spanish, German, Turkish, Hindi, and more
Hakim Arab v2 — Dialect-aware STT with 90ms latency, speaker diarization, and up to 60-minute audio per request

Drop-in OpenAI-compatible endpoints (/v1/audio/speech and /v1/audio/transcriptions), so you can swap in 5 lines of code. Data stays in the region you choose — UAE, Saudi Arabia, Qatar, or Frankfurt.

Free tier included. No card required. First audio sample in 30 seconds.

👉 tryhakim.ai

1d:T5a5,

Hakim AI — Arabic-First Voice AI for Developers & Creators

The speech intelligence layer the Arabic internet has been waiting for.

Hakim gives developers and creators state-of-the-art TTS, STT, and voice cloning — built natively for Arabic, not retrofitted from English. Most voice APIs treat Arabic as an afterthought. Hakim was designed around it from day one: 15 Arabic dialects (Gulf, Egyptian, Levantine, Moroccan, Iraqi, and more), native diacritics handling, right-to-left text, and Arabic-English code-switching in the same sentence.

Two models, one API:

Hakim Fast v1.3 — Streaming TTS with 113ms time-to-first-byte, 15 Arabic dialects, and 30+ languages including English, French, Spanish, German, Turkish, Hindi, and more
Hakim Arab v2 — Dialect-aware STT with 90ms latency, speaker diarization, and up to 60-minute audio per request

Drop-i

Hakim AI

About