
HUMAIN Voice
Arabic-first speech APIs: dialect-aware speech to text, text to speech and voice agents.

About HUMAIN Voice
HUMAIN Voice: Arabic-First Speech Recognition, Text to Speech and Voice Agents
HUMAIN Voice is an Arabic-first voice AI platform with three APIs: speech recognition, text to speech and speech to speech for voice agents. It is built for Arabic as people actually speak it, across dialects and with Arabic-English code-switching, and every API works with the same key.
Key Features
Dialects and Code-Switching: Arabic and English with automatic dialect detection and mid-sentence switching between the two. HUMAIN says more than 15 dialects are in production, including Saudi, Khaleeji, Egyptian, Levantine and Maghrebi variants.
Three Transcription Modes: Batch for complete recordings, Fast for one short clip such as a turn in a conversation, and Realtime for live 16 kHz audio with partial and final transcripts.
Batch Options: Speaker diarization at three levels, redaction of personal data, and number and date formatting.
Telephony Model: A dedicated Arabic model for 8 kHz phone-call recordings, available in batch.
Text to Speech: Seven stock voices (four Saudi, one Levantine, one Moroccan, one Egyptian), all speaking Arabic and English, with styles such as news, storytelling, advertising and call center, and emotions including happy, sad, angry, fearful and whisper.
Voice Cloning: Adapt a voice from a reference clip and its transcript. Cloning requires a signed-in account.
Voice Agent (Beta): A hosted agent that listens and answers in Arabic and English. Developers set the prompt, intro, voice and client tools, users join from a browser, or from iOS, Android, Flutter and React Native apps through LiveKit clients. The reply dialect is automatic or fixed.
SDKs and APIs: TypeScript and Python SDKs with the same call shape, Go announced, plus REST and Socket.IO APIs.
Data Residency: Compute and storage run in Saudi Arabia, with private deployment options for regulated industries.
Use Cases
Contact Centers: Transcription of Arabic phone calls and voice agents that reply in the caller's dialect.
Captions and Meetings: Live captions, subtitles and meeting transcripts with speaker labels.
Voice-overs and IVR: Arabic narration, IVR prompts and read-aloud content.
Industry Assistants: Voice assistants for banking, healthcare, real estate, retail and food and beverage, the presets in the demo.
Pricing
Free: 5K credits with no card, about 3 hours of batch transcription.
Monthly plans from 49 SAR (Starter, 75K credits) to 649 SAR (Scale, 1.29M credits), plus custom Enterprise plans.
Pay as you go: 1.00 SAR per audio hour for batch transcription, 1.80 SAR for realtime, 0.224 SAR per 1,000 characters of speech and 0.51 SAR per voice agent minute.
Getting Started
Website: https://voice.humain.com/
Documentation: https://docs.voice.humain.com/
HUMAIN Voice covers Arabic transcription, speech synthesis and voice agents from one API key, with a free tier to start.
