Shillong-based MWire Labs has released Lemka, a voice AI speech system supporting six languages of Northeast India.
DIMAPUR — MWire Labs, a research and AI deployment lab based in Shillong, has released Lemka, an end-to-end speech AI system that combines speech-to-text (Lemka STT) and text-to-speech (Lemka TTS) for six languages of Northeast India: Nagamese, Garo, Mizo, Meitei (Manipuri), Kokborok, and Khasi.
According to an update, Lemka allows users to speak in their own language and be understood by a computer system, and for that system to reply in the same language without requiring typing, reading, or knowledge of English or Hindi.
The technology is designed to support real-world use cases, such as citizens accessing government helplines and welfare schemes in their native language, farmers receiving agricultural advisories aloud in the field, and students interacting with learning tools that speak directly to them.
The system is built on more than 1,000 hours of proprietary field audio recorded across the region and is architected for practical deployment, including offline and edge-based settings suited to low-connectivity environments. This builds on the lab's earlier published work on an offline voice assistant for Garo, which was developed for visually impaired users and runs entirely without cloud dependency.
It was claimed that the models are optimised for real-world noisy and dialectal speech common in the region.
Lemka follows MWire Labs’ earlier open release of NE-ASR, a speech recognition model covering eight Northeast Indian languages, on Hugging Face. It forms part of the lab’s broader NE-Stack, a foundational language AI infrastructure being built for the region’s more than 220 indigenous languages, most of which currently have no AI support.
Lemka is currently available for deployments and early access partnerships. MWire Labs is currently engaging government departments, research institutions, and organisations in healthcare and agriculture for early pilot deployments while also contributing to collaborative language technology efforts for Northeast Indian languages.
Director of MWire Labs Badal Nyalang stated, “Our languages carry our history, culture, and identity. With Lemka, we’re not just building speech models, we’re making sure Northeast India’s voices are actively woven into the future of technology, rather than left behind.”