Gnani AI Releases Prisma v3.0 Speech Recognition Model Covering 10 Languages

Gnani AI has launched Prisma v3.0, a speech recognition model that supports nine Indian languages plus English. The company says the model achieves lower transcription error rates than competing models on a multilingual benchmark. This matters for businesses and developers who need accurate voice interfaces for India's diverse linguistic market. By covering languages such as Hindi, Tamil, Telugu, and Bengali, Gnani AI aims to make voice technology more accessible to millions of users. The lower error rates could improve customer service bots, voice assistants, and transcription tools. Gnani AI did not disclose the exact error rate figures or the specific benchmark used. However, the launch signals growing competition in Indian speech AI, where local players are challenging global tech giants. The company plans to expand language support further, potentially adding more regional dialects. For enterprises targeting Indian consumers, Prisma v3.0 offers a way to build voice applications that understand local speech patterns more reliably.