Technology
1 min read
Shunya Labs launched Zero Codeswitch, a foundation model achieving a 3.10% Word Error Rate on the Op...
The AMW Read
Shunya Labs introduces a specialized foundation model for code-mixed speech, advancing the sovereign AI/linguistic diversity pattern (Frame 2) and signaling a shift toward localized, cost-efficient model deployment.
NoveltySignificance
Foundation Models Β· Player MapGeopolitics
Shunya Labs launched Zero Codeswitch, a foundation model achieving a 3.10% Word Error Rate on the OpenASR leaderboard, a 48% improvement over competitors. It natively processes code-mixed speech like Hinglish, trained from scratch on millions of hours of local data. Its CPU-first design slashes deployment costs by 20x while delivering sub-100ms latency for enterprise scale. This marks a major shift toward cost-efficient, sovereign AI tailored for India's linguistic complexity. π
