PreVox™ Neural NRT is ready-to-deploy noise cancellation and speech enhancement software, custom-trained for your noise conditions.
- Offline (Non-Real-Time) processing for maximum intelligibility with preserved timbre.
- Delivered as a library for Linux, Windows, and macOS and it can be ported to other platforms that support modern deep-learning runtimes.
- Model trained on your sample recordings and handles stationary and non-stationary noise, including reverberation.
- Higher quality than real-time: full-context processing with larger models.
- Minimal artifacts: no “musical noise,” natural prosody retained.
- Optional objective reporting: ESTOI/STOI and PESQ.
Please contact us for details.