CamMicGo Docs
Microphone test documentation
A detailed guide to raw vs call processing, waveform, frequency spectrum, dBFS, background scenarios, A/B recordings and quality scoring.
Raw microphone vs call processing
Raw mode asks the browser to disable echo cancellation, noise suppression and automatic gain control. Call processing asks for those three features. The browser or operating system may still apply hardware or driver processing, so “raw” means minimally processed by the browser rather than a bit-perfect hardware feed.
Waveform and frequency spectrum
The waveform shows amplitude over time: it is useful for spotting silence, speech dynamics and clipping. The frequency spectrum uses an FFT to show how energy is distributed from low to high frequencies. Strong low-frequency energy may indicate rumble or hum; a broad voice spectrum generally extends through the midrange with consonant detail in the upper bands.
dBFS, noise floor and voice/noise margin
CamMicGo reports digital level in dBFS. 0 dBFS is the digital ceiling; values closer to 0 are louder and clipping can occur near that ceiling. The quiet-room measurement estimates a noise floor, then the voice/noise margin compares your current speech level with that floor. These values are relative digital measurements and are not calibrated dB SPL sound-pressure measurements.
Background scenarios and A/B recordings
Background ambience can be used for listening tests and can also be included in the saved local A/B recording when you select it for that slot. This makes the result reproducible: the recording you listen back to contains the chosen test context. For a true test of microphone or browser noise suppression, play ambience through speakers so it physically reaches the microphone; a digitally mixed background is useful for voice-over-background comparison but does not pass through the microphone’s capture processing.
How the quality score works
The indicative score combines recorded volume, signal clarity, clipping, frequency coverage, dynamic range and, only when a background was actually included, background/noise handling. A metric that was not tested is shown as “Not evaluated” and receives no hidden 100/100. Its weight is removed and the remaining metrics are renormalised. This avoids rewarding a microphone for a capability that was never exercised.
Why two sites can show different numbers
Online microphone tools can use different RMS windows, silence percentiles, clipping thresholds, FFT sizes, smoothing and score weights. Two tools can therefore reach a similar overall grade while showing different sub-metrics. CamMicGo treats its result as a repeatable browser diagnostic rather than a calibrated laboratory certificate.