JumpML Noise Reduction
Problem and motivation
- Noisy World
- My Life
- Going on mute
The only constant in life is change... and background noise.
Our Solution
- JumpML Noise Reduction powered by SignalSifter
-
Cut the background noise and preserve the speech, so you are heard clearly
Product benefits
- Strong Performance
- Cuts diverse noises
- Improve intelligibility
- Improve MOS/WER
- SignalSifter optimized for audio
- Energy-efficient
- Made for embedded
- Operates at 16kHz
- Uses efficient RNN
- Low latency
- 10 ms or less
- Useful for hearing aids
- Smart transparency
Forget about the noise and speak with ease, anywhere and anytime.
Comparison Table
| Features | JumpML NR | RNNoise | Other offerings |
|---|---|---|---|
| Good noise reduction | Yes | Depends | Yes |
| Model/Memory < 0.4 MB | Yes | Yes | No |
| Low-latency <= 10 ms | Yes | Yes | No |
| Real-time on tiny embedded | Yes | Yes | No |
| Privacy (no internet needed) | Yes | Yes | No |
| Portable/inspectable C code | Yes | Yes | No |
| Adjustable Noise Reduction | Yes | Easy | No |
| Customizable | Yes | No | No |
More Info
For JumpML Noise Reduction audio demos and specific use cases, please checkout our Use Cases page.
For questions about how it can fit into your product, or about what customizations are available for your use case, or if you just want more details about the implementation, or just to say hello, please contact us at JumpML email.