Why use Nyxhora for this workflow
RNNoise-based AI denoising removes keyboard clicks, fan noise, echo, and background chatter in real-time
Toggle noise reduction per-recording or per-clip — keep the flexibility to compare denoised vs original
Runs in-browser with no audio file uploads to external servers — your audio stays on your device
Why screen recordings have audio problems (and how to fix them)
Screen recordings are captured in uncontrolled environments — home offices, coffee shops, open-plan workspaces. The microphone picks up everything: keyboard clacking, mouse clicking, air conditioning hum, fan noise, traffic outside, room echo (sound bouncing off walls), and background conversations. These sounds are subtle in person but amplified in recordings because: (1) Microphones are sensitive and close to your keyboard. (2) Compression algorithms (used in MP4 encoding) amplify background noise. (3) Headphone listeners hear everything more clearly than speakers. The fix: AI noise removal using RNNoise. Unlike simple noise gates (which mute quiet passages), RNNoise analyzes the frequency spectrum and distinguishes voice from non-voice sounds. It preserves natural voice quality while suppressing specific noise types: keyboard typing, fan hum, room echo, and background chatter.
Studio Audio vs post-processing: which approach is better
Two ways to apply noise removal: REAL-TIME (Studio Audio during recording): The AI processes your audio as you record. The denoised audio is captured directly. Benefit: you get clean audio from the start. No post-processing needed. Tradeoff: slight latency in monitoring (you hear the processed audio, not the raw input). POST-PROCESSING (apply in editor): Record normally, then apply denoising to the clip in the editor. Benefit: you can compare denoised vs original and adjust. Tradeoff: extra step, slight quality loss from double-processing. RECOMMENDATION: Use Studio Audio for new recordings. Use post-processing for existing recordings that were made without noise removal. Do NOT use both on the same recording — double-processing can introduce artifacts.
What AI noise removal can and cannot do
WHAT IT CAN REMOVE: keyboard typing sounds, mouse clicks, fan and AC hum, room echo/reverb, background conversations (if quieter than your voice), traffic noise (if distant). WHAT IT CANNOT REMOVE: your voice (by design — it preserves speech), sounds louder than your voice (construction next door), music playing in the room, multiple people talking at similar volumes. WHAT IT STRUGGLES WITH: heavy accents combined with noise (the model may confuse accent features with noise), very fast speech (less data per word for the model to work with), extremely noisy environments (better to fix the environment than rely on AI). The practical approach: AI noise removal is excellent for typical screen recording noise. For extreme cases, fix the source first (close windows, turn off fans, move away from noise) then apply AI for residual cleanup.
Audio quality checklist for screen recordings
BEFORE RECORDING: (1) Close unnecessary applications (reduces fan noise). (2) Close windows (reduces traffic/outdoor noise). (3) Turn off AC if possible (or point mic away from vent). (4) Silence your phone. (5) Sit 12-18 inches from your mic. (6) Enable Studio Audio. DURING RECORDING: (1) Speak clearly at consistent volume. (2) Avoid tapping the desk or rustling papers. (3) Type gently if you must type during recording. AFTER RECORDING: (1) Listen back with headphones — you will catch issues you missed. (2) Check for consistent volume levels. (3) Apply noise removal if not using Studio Audio. (4) Adjust clip volume if needed (voice should peak around -6dB to -3dB). This checklist takes 2 minutes and saves hours of post-processing.
How it works
Record with Studio Audio enabled
In the Nyxhora recorder, toggle Studio Audio before recording. This applies real-time AI noise reduction during capture. The cleaned audio is recorded directly.
Or apply to existing clips
In the editor, select an audio or video clip and toggle the Studio Audio denoise option in the inspector. Useful for recordings made without noise removal.
Export with clean audio
Export your video as normal. The denoised audio is baked into the final output. No separate audio processing step needed.
Frequently asked questions
Does noise removal affect voice quality?
RNNoise is specifically trained to preserve voice quality while removing noise. It maintains natural tone, clarity, and warmth. Double-processing (applying twice) can degrade quality — use once per recording.
Is AI noise removal a Pro feature?
Studio Audio (real-time during recording) is available to all users. Advanced per-clip controls and batch processing are available to Pro users.
Can I adjust the intensity?
The RNNoise model has fixed sensitivity optimized for voice. Default settings work well for most recordings. For extreme cases, fix the source noise first rather than relying on higher intensity.
Does it work with multiple speakers?
RNNoise preserves all speech while removing non-speech sounds. It works with multiple speakers as long as the noise is distinctly non-vocal (fans, keyboards, echo).
Related guides
Explore more workflows and features in the Nyxhora Creator Studio.