How to find out whether the hiss was in your recording or added by the edit
Give Vizard Agent the raw microphone files as well as the finished video and ask it to measure the noise floor in the pauses of each. If the raw is quiet in its silences and the export is not, the hiss was manufactured somewhere in the processing. That needs a different repair from cleaning a noisy recording.
What do you type into Vizard Agent?
Ask for the diagnosis before the cure. "Clean up this hiss" gets you a denoised file whether or not denoising was the right answer, and a denoised file always sounds slightly worse than the original. Give Vizard Agent the raw microphone files too, because without them there is nothing to compare the export against.
Prompt
What does Vizard Agent actually do?
A hiss in the delivered file has two possible origins, and they need opposite fixes. Denoising a clean recording damages the speech for nothing, while rebuilding the chain fixes nothing if the microphone really was noisy. So almost every step is measurement, and the cleaning is one decision at the end. In the session behind this article it:
- Located silent passages in the raw material and measured the noise floor in them on each microphone, then measured the same on the edited episode.
- Viewed the spectrogram of a silent passage on both microphones and scanned both across the episode for bursts of static.
- Produced a spectrogram of the suspect passage in the export and compared the spectra of speech, music and raw microphone at that point.
- Checked how the export's speech track had actually been built, since gain staging, an aggressive processor or a bleeding second mic can add a floor that was never recorded.
- Rebuilt the speech track from the raw microphones and compared, then tried several cleaners, listened to each and picked the most natural.
When does this not work well?
Some recordings have no usable silence. Continuous music, a live room, or a speaker who never pauses for breath leaves Vizard Agent nowhere to measure a floor, and the comparison the method depends on gets much harder. Without the raw files no comparison is possible, and intermittent bursts won't show up in a floor measurement.
Common questions
Why do I need the raw files? Because the comparison is the diagnosis. Without them Vizard Agent can only clean.
Can processing really add hiss? Yes. Vizard Agent checks gain staging, aggressive processors and microphone bleed.
What if only one microphone is bad? Vizard Agent measures them separately and can rebuild from the clean one.