Enterprise Voice AI platform for Speech-to-Text, Text-to-Speech, and Voice Agents
High-assurance infrastructure for AI agents
Stage
Established
Not given
Founded
2015
Not given
Based in
US
US
Pricing model
Freemium
Open source
Pricing
Flux STT Streaming: $0.0065/min (Pay As You Go), $0.0057/min (Growth); Flux Multilingual Streaming: $0.0078/min (Pay As You Go), $0.0068/min (Growth); Nova-3 Monolingual Streaming: $0.0048/min (Pay As You Go), $0.0042/min (Growth); Nova-3 Multilingual Streaming: $0.0058/min (Pay As You Go), $0.0050/
Not given
Key features
Flux STT with turn detection and interruption handling
Nova-3 multilingual speech recognition in 50+ languages
Ultra-low latency under 300 milliseconds
Keyterm prompting with up to 90% higher keyword recall
Flux TTS with latency as low as 80ms
Speaker diarization and multi-speaker audio detection
Smart formatting with automatic punctuation and capitalization
Automatic redaction of sensitive information
Custom models for edge-case scenarios
Flexible deployment options: cloud, in-VPC, or self-hosted
Harpe: chooses what an AI agent reads under a token budget with consistent results
Kibisis: keeps agent memory on your hardware with AES-256-GCM encryption
Argus: append-only record that detects edits with cryptographic verification
No network code by default for local and disconnected use
MIT License with published code, benchmarks and raw results
What makes it different
Conversation-aware speech platform with native turn-taking detection and interruption handling
Unified single API combining STT, TTS, and LLM orchestration to reduce complexity and latency
Flux TTS that thinks before speaking to maintain tone and expression across conversations
Enterprise-grade accuracy with custom vocabulary injection and industry-tuned models
Foundational AI company with research-driven innovation and continuous model improvements
Measured empirical validation across 6,111 questions outperforming full-context models
Zero plaintext on disk with encrypted vaults matching plaintext accuracy
78.3% fewer tokens sent to the model with a 1,200-token budget per turn
Preserves conversational chronology and relational qualifiers under token budgets
Deepgram
Perseus Computing