Hush
Weya’s AI noise suppression tool that removes background noise from calls and recordings, delivering clear audio with Hush v1.0.
What is Hush?
Hush is an in-house open-source noise suppression model developed by Weya AI. It strips background noise and competing voices from real-world calls, turning chaotic audio into clean, ASR-ready speech. Users deploy Hush to improve call clarity for Voice AI applications, ensuring bots, agents, and compliance teams get better outcomes from clearer audio signals.Application scenarios
Busy cafes
Removes distracting sounds like conversations, machines, or clattering dishes.
City streets
Handles traffic noise to keep calls understandable.
Construction sites
Suppresses heavy background noise from tools and equipment.
Office environments
Filters out office buzz and overlapping speaker chatter.
Live agent calls
Keeps agents hearing customers clearly during real-time conversations.
Voice AI pipelines
Cleans call audio before it reaches ASR systems for better transcription accuracy.
Core Features
Real-time CPU processing
Cleans each 10 ms of audio in under 1 ms on standard CPUs, with no GPU required.
Voice isolation
Focuses on the main caller and pushes background talk, TV noise, and competing voices aside.
Noise handling for tough environments
Handles traffic, office buzz, fans, and street sounds effectively.
Lightweight model
Approximately 8 MB in size, making it easy to deploy in your own cloud or data center.
Low latency
Typical processing time of 0.9 ms per 10 ms audio frame, keeping calls real-time.
Open-source availability
Available on GitHub for community use and customization.
High-quality training data
Trained on over 10,000 hours of real-world noisy audio, including overlapping speakers and tough environments.
Top-ranked performance
Ranked #5 on Hugging Face’s Audio-to-Audio leaderboard at launch.
Target users
Hush is designed for Voice AI developers, call center operators, BFSI teams (banking, financial services, and insurance), and any team building voice bots or agent-assist systems. It also benefits compliance teams who need clean audio for accurate transcription and analysis.How to use Hush?
Visit the official Hush page at https://www.weya.ai/hush and click "Try Hush Now" to test the model on live calls. For integration, download the open-source model from GitHub and deploy it in your own cloud or data center environment. No special hardware is required—it runs on standard CPUs.Pricing and free trial
The website text does not mention any pricing details or free trial options. Visit the official site for current pricing information.Effect review
Hush delivers on its promise of turning noisy call audio into clean speech, backed by a top-5 ranking on Hugging Face’s leaderboard. The 0.9 ms processing time on CPU ensures calls stay real-time without needing expensive GPU hardware. With 10,000+ hours of real-world training data, the model handles overlapping speakers and tough environments effectively. For teams building Voice AI systems, Hush provides a lightweight, open-source foundation that fixes audio quality at the source—reducing ASR errors and improving overall call outcomes.Frequently Asked Questions
What is Hush?
How does Hush work?
What platforms does Hush support?
Is Hush free to use?
Does Hush work offline?
Hush - AI Tool Detail
Weya’s AI noise suppression tool that removes background noise from calls and recordings, delivering clear audio with Hush v1.0.
Category：Speech processing
Visit Link：https://www.weya.ai/hush
Tags：AI noise suppression、background noise removal、audio enhancement、call clarity、noise cancellation