Hush
Hush transforms noisy phone conversations into clear audio suitable for voice AI applications.
Open SourceAudio & Music
Visit website Free to browse · updated 2026-06-23
What is Hush?
The model isolates the primary speaker's voice while reducing distractions such as traffic, office activity, or cafe chatter. This approach ensures that automatic speech recognition systems receive higher quality input, leading to fewer interruptions and more accurate transcriptions during conversations. Hush integrates directly into existing voice AI workflows without requiring specialized hardware. It supports use cases in sectors like lending and insurance by maintaining call clarity even in challenging acoustic conditions. By focusing on practical deployment and low latency, the tool helps organizations achieve more reliable outcomes from their voice-based automation and compliance processes.
Key features
Open-source noise suppression model for Voice AI calls
Removes background noise and competing voices in real time
Processes each 10 ms audio frame in 0.9 ms on CPU
8 MB model size for easy deployment
Trained on 10,000+ hours of real-world noisy audio
Ranks #5 on Hugging Face Audio-to-Audio leaderboard
Isolates main caller voice while handling traffic, office, cafe, and street noise
AI models Hush uses
Hush v1.0
in-house open-source noise suppression model
What you can use Hush for
Debt Collection
Hush improves debt collection calls by stripping background noise and competing voices so agents and Voice AI systems hear the caller clearly in real time.
Insurance
Hush delivers clean, ASR-ready audio for insurance workflows, handling office buzz, traffic, and street sounds to reduce repeat requests during live calls.
Loan Sanction
Hush isolates the main caller voice in loan sanction conversations, enabling reliable Voice AI performance even in noisy cafe or construction environments.
How to use Hush
- 1Access the open-source model on GitHub
- 2Download the lightweight 8 MB model file
- 3Integrate Hush into your existing Voice AI call pipeline
- 4Feed live audio frames for real-time CPU processing
- 5Monitor output for isolated caller voice and reduced noise
Hush pricing
Pricing model: Open Source. Plan details are indicative — check the site for current prices.
Open Source
Free
- Github access
- Real-time CPU inference
- 8 MB model size
Editor's verdict
Pros
- +Real-time CPU processing with no GPU required
- +Lightweight 8 MB model for cloud or on-prem deployment
- +Open-source with GitHub availability
Cons
- –Pricing details unavailable on site
- –No mobile or browser extension support mentioned
Our take: Hush is a solid audio & music choice. It's valued for real-time cpu processing with no gpu required and lightweight 8 mb model for cloud or on-prem deployment. The main trade-off is pricing details unavailable on site. A good pick if you want capable AI without a high upfront cost.
Frequently asked questions
Hush is weya AI’s open-source noise suppression model that removes background noise and competing voices from Voice AI calls in real time.
Summary
Hush is a solid audio & music choice. It's valued for real-time cpu processing with no gpu required and lightweight 8 mb model for cloud or on-prem deployment. The main trade-off is pricing details unavailable on site. A good pick if you want capable AI without a high upfront cost.
Did you find this helpful?
User reviews
Verified reviews from the community shape this tool's rating.
Loading reviews…
Hush alternatives
Similar audio & music tools worth comparing.
