Show HN: Sokuji – speech real-time translation powered by Gemini Live APIs

Sokuji Now Supports Gemini 2.5 Flash & 2.0 Flash Live APIs (Beta)

Real-time AI translation powered by Google's latest Gemini models - now in beta testing! We're excited to announce that Sokuji now supports Google's cutting-edge Gemini APIs in beta:

New Beta Features: - Gemini 2.5 Flash Live API Native Audio Preview - Advanced audio processing, optimized for Asian languages (Japanese, Korean, etc.) - Gemini 2.0 Flash Live API Preview - Cost-effective multimodal AI, excellent for major languages (English, French, Portuguese, etc.) Note: This is currently in experimental phase as we integrate these powerful new APIs.

Free Tier Available: Gemini 2.0 Flash (More cost-effective): - 15 requests/minute, 1M tokens/minute, 1,500 requests/day - Great for: English, French, Portuguese, Spanish, German Gemini 2.5 Flash (Enhanced language support): - 10 requests/minute, 250K tokens/minute, 500 requests/day - Optimized for: Japanese, Korean, Chinese, and other Asian languages

Live API Native Audio: - 1-3 concurrent sessions, 25K-50K tokens/minute Source: Google AI Rate Limits Documentation https://ai.google.dev/gemini-api/docs/rate-limits

Choose Your Model: Budget-conscious + Major languages → Gemini 2.0 Flash Asian languages + Premium features → Gemini 2.5 Flash Both available for free during beta testing!

Beta Status: We're actively testing and refining the integration. Some advanced features are still being implemented. Your feedback helps us improve!


Comments URL: https://news.ycombinator.com/item?id=44324883

Points: 1

# Comments: 0