# ai-coustics

> B2B audio infrastructure for voice AI. Provides real-time speech enhancement, voice isolation,
> and audio preprocessing via on-device SDK. Designed to run before VAD and ASR in voice agent
> pipelines (LiveKit, PipeCat, custom stacks) to improve transcription accuracy, reduce latency,
> and stabilize turn-taking.

## Products
- [Quail](https://ai-coustics.com/quail): Real-time enhancement optimized for machines (ASR + VAD). Reduces noise, isolates foreground speaker (Voice Focus), improves WER. <30ms latency, CPU-only, language-agnostic.
- [Rook](https://ai-coustics.com/rook): Enhancement optimized for human listening (call recordings, media).
- [Benchmarks](https://ai-coustics.com/benchmarks): Performance comparisons across noise conditions, ASR accuracy, and latency.

## Integration
- SDK: C++, Rust, Python, Node.js, WebAssembly
- Drop-in layer before VAD and ASR in existing pipelines
- Compatible with LiveKit (cloud + self-hosted), PipeCat, custom stacks
- Runs fully on-device — no audio leaves the system

## Developer Resources
- [SDK Tutorial](https://ai-coustics.com/blog/ai-coustics-sdk-tutorial-four-steps): Step-by-step integration guide
- [SDK for Voice Agents](https://ai-coustics.com/blog/sdk-fixing-audio-input-for-voice-agents): Improving voice agent input quality

## Company
- [About](https://ai-coustics.com/about)
- [Pricing](https://ai-coustics.com/pricing)
- [Book a Demo](https://ai-coustics.com/book-demo)
- [Contact](https://ai-coustics.com/contact)
- [Blog](https://ai-coustics.com/blog)
- [Voice AI Glossary](https://ai-coustics.com/glossary)

## Legal
- [Privacy Policy](https://ai-coustics.com/legal/privacy-policy)
- [Terms of Service](https://ai-coustics.com/legal/terms)
- [Legal Notice](https://ai-coustics.com/legal/legal-notice)