ai-coustics
Real-time speech enhancement SDK that cleans audio for voice AI agents and STT
ai-coustics is an AI tool for music. It is available on API, Self-hosted, Linux, macOS and Windows.
- Pricing
- Contact sales
- Platforms
- API, Self-hosted, Linux
- Listing updated
Similar tools
About ai-coustics
ai-coustics builds audio models that clean up speech before it reaches speech-to-text, voice activity detection and language models. The company's SDK isolates the foreground speaker, suppresses background noise and competing voices, and processes audio in under 10 ms, with the site citing about 30 ms latency for its Quail models at 8 and 16 kHz PCM.
The product is packaged as a lightweight SDK that the site says needs no GPU and no ONNX dependency, and it is offered with on-prem deployment on higher tiers. Its model line includes Quail for speech enhancement, VAD for voice activity detection and Tyto for predicting whether degraded audio will cause downstream failures in an agent.
The site targets voice agent builders, telephony teams and voice cloning or creator-tool developers. Published case studies from PolyAI, telli, Synthesia and Elgato describe reduced false barge-ins and fewer audio-related failures, though those results are reported by the customers and vendor.
What it does
- Real-time speech isolation
- Suppresses background noise and competing voices to keep the foreground speaker clear.
- Voice activity detection
- A VAD model designed to work without a separate de-noising step.
- Audio decision scoring
- Tyto scores whether incoming audio is likely to cause downstream failures and gives the cause.
- Low-latency SDK
- Quoted at about 30 ms latency at 8 and 16 kHz PCM, running on CPU without a GPU.
- On-prem deployment
- Higher plans include on-premises SDK deployment and offline or air-gapped license options.
Good for
- Improving speech-to-text accuracy for noisy inbound phone calls
- Reducing false barge-ins in voice agents
- Cleaning upstream audio for voice cloning and avatar training
- Stabilizing voice activity detection in busy environments
Pricing
Contact sales, from $135 a month
Monthly plans from $135 (annual billing); Enterprise starts at $2000/month.
Prices change. Check ai-coustics.com before you buy.
ai-coustics FAQ
Can I test before subscribing?
The site offers a free try option and a Developer Platform where you can test models and generate SDK keys.
What happens if I go over the included minutes?
The pricing page lists per-minute overage rates for each plan, from 0.15 cents per minute on Startup down to 0.11 cents on Scale.
Does ai-coustics support on-premises deployment?
Yes, the Startup plan lists on-prem SDK deployment, and the Enterprise plan includes offline or air-gapped license options.
How much does ai-coustics cost?
ai-coustics does not publish prices. Contact the company for a quote.
What are the best alternatives to ai-coustics?
The closest alternatives to ai-coustics are Audo AI, Cleanvoice AI, CrystalSound, Supertone and Aflorithmic.
Where this comes from
- Description: written from the project's website.