Curated by
More in The AI OS and Context Graph
See all 6 →Commotion: AI That Actually Works
gocommotion.com
The Commotion AI OS: Connecting Context to Execution
gocommotion.com
Context Graphs Explained: The Foundation of Enterprise AI
synlabs.io
More from Commotion
See all stacks →Voice AI 101
7 cards · Technology
AI Governance, Security and Compliance
10 cards · Technology
Moving AI from Pilots to Production
8 cards · Technology
Commotion LinkedIn: Company Updates and Product News
Commotion Laya v1.5 voice AI model outperformed competing models (Cartesia, Sarvam, ElevenLabs) in blind human evaluations for Indian-language customer support telephony, achieving 64.6-62.9% overall preference and 86-92% preference for empathy in sensitive scenarios. The benchmark involved 332 evaluators, 10,500+ votes across 8 languages and 9 industries, with Laya demonstrating superior performance through lower error rates in pacing, pronunciation, and word delivery.
Built for AI agentsACO · 348 tokens
Summary
Commotion Laya v1.5 voice AI model outperformed competing models (Cartesia, Sarvam, ElevenLabs) in blind human evaluations for Indian-language customer support telephony, achieving 64.6-62.9% overall preference and 86-92% preference for empathy in sensitive scenarios. The benchmark involved 332 evaluators, 10,500+ votes across 8 languages and 9 industries, with Laya demonstrating superior performance through lower error rates in pacing, pronunciation, and word delivery.
Tags
voice-ai · customer-support · benchmark · telephony · indian-languages · laya-v1.5 · empathy
Key entities
Commotion (organization, 0.95) · Commotion Laya v1.5 (technology, 0.95) · Cartesia Sonic v3 (technology, 0.9) · Sarvam Bulbul v3 (technology, 0.9) · ElevenLabs Turbo 2.5 (technology, 0.9) · Josh Talks (organization, 0.85) · voice-AI-benchmark (concept, 0.92) · Indian languages (location, 0.88)
Classification
analysis · language en · status final
Provenance
claude-haiku-4-5 via @stacklist/be@0.1.0, confidence 0.85, 8 Jun 2026