Saudi Arabia’s HUMAIN has launched HUMAIN Voice, a speech AI platform designed around Arabic dialects and the way people naturally switch between Arabic and English. The platform combines automatic speech recognition, text-to-speech and voice agents.
HUMAIN says the system supports Saudi dialects including Najdi and Hijazi, as well as Gulf and other Arabic dialects, English and Arabic-English code-switching. Its voice agents are designed to listen and respond in natural, full-duplex conversations, including replies in a caller’s dialect. The company sees potential uses in customer service and enterprise operations.
Alongside the platform, HUMAIN is releasing HUMAIN-Voice-Lite-80M, an open-weight automatic speech recognition model with 80 million parameters. Released under the Apache 2.0 license, it is optimized for CPU and GPU environments, which HUMAIN says could make it useful for applications where cost, latency or on-device deployment matters.
HUMAIN says it will publish benchmark results for the model on the Open Universal Arabic ASR Leaderboard, along with a blog about its performance across Arabic, English and code-switching speech scenarios. The company’s VP of HUMAIN Voice, Dr. Ahmed Ali, said the models treat dialect, accent, context and code-switching as core features of speech rather than exceptions.
HUMAIN is a PIF company that works across AI data centers, infrastructure and cloud platforms, AI models and applied solutions. CEO Tareq Amin said HUMAIN Voice is part of the company’s effort to build AI interfaces that let people communicate naturally, rather than adapting how they speak to technology.
Source: Zawya AI


