Signal-to-noise ratio (SNR)
How far the voice or music sits above the hiss, hum, and room noise. A bigger dB gap means a cleaner recording.
See it
What it is
Signal-to-noise ratio compares the wanted sound with the unwanted background, expressed in decibels. Higher is cleaner. If speech averages -20 dBFS RMS and a matching room-tone section averages -60 dBFS RMS, the recording has a 40 dB SNR for those measurements.
Reach for it while choosing a room, microphone position, preamp, or recording level. Move the mic closer to the voice, turn off the fan, and aim a directional mic away from traffic before recording. Those moves increase the wanted signal relative to the noise; raising a finished file does not. Cassette tape managed roughly 50 to 60 dB, which is why Dolby B and dbx noise reduction existed at all. On a 24-bit recorder the converter is no longer the weak link: the room, the mic, and the preamp are.
Gotcha: SNR depends on what you call signal, which noise interval you sample, and the bandwidth or weighting of the measurement. A single number can also hide a very audible 50 or 60 Hz hum. Normalization and makeup gain lift signal and noise together, leaving the ratio unchanged.
Ask AI for it
Use FFmpeg's astats filter to measure RMS level in one clean speech selection and one nearby room-tone selection from the same recording. Subtract the room-tone RMS dBFS from the speech RMS dBFS to report SNR in dB, list both source intervals with timestamps, and flag any narrow 50 or 60 Hz hum separately from the broadband noise figure.