Gnani.ai Doubles Down On Sovereign Voice AI Models With Prisma v2.5 Launch

by · Inc42

SUMMARY

  • Gnani.ai has launched Prisma v2.5, a speech-to-text model spanning 12 languages and trained on 14 Mn hours of proprietary Indic speech data.
  • The startup claims the model delivers higher accuracy and lower latency for real-world Indian voice environments, including multilingual and noisy telephony conversations.
  • The launch comes as Indian sovereign AI players intensify competition with global voice AI firms such as ElevenLabs and Wispr Flow.