[paddington]

Let me stop you guys from rolling down the wrong track with this FM radio and 16kHz thing. What Audioarchivist is thinking about in terms of the high frequency help in FM transmission is a pre-emphasis curve of 75uS (microseconds) - which raises the modulation of the higher frequencies in a logarithmic curve that is then de-emphasised in the receivers. This allows a better S/N ratio over the entire system to combat the noise that is naturally more prevalent in the higher end of the audio spectrum.

If you record from the receiver, it should already be de-emphasised, and you wouldn't see much in the analysis to betray the presence of the processing (unless the receiver is a piece of shit). FM transmission has a response of 15kHz on each channel and rolls off sharply after that. That doesn't mean you would see a tone at 16kHz. I don't know where people get that from.

Now, you may still see the 19kHz stereo pilot tone. This tone tells the receiver to 'de-mux' the audio it gets to 2 stereo channels using the info at the second harmonic up from the pilot (38kHz). This is done to allow the same FM transmission to be received on both mono and stereo receivers. The L-R signal, or 'difference' signal is applied to the sum signal (L+R) and you get "stereo separation" to whatever degree the difference (L-R) signal is applied. In most cases, you get about 20% or so of the L-R signal. The people at Dolby stole this method when they created the original 4 channel Sourround Sound, for the back channel. It's the L-R difference signal in the mux.


Anyway.. the 16kHz noise needs to be analyzed before we can decide what it is. Don't assume it's a tone (which assumes a sine wave) - I'm betting it's white noise.

When I say "quiet" spot, I don't mean "silent" - any quiet passages in the music will be fine. The reason I ask for quiet is that the noise is likely a constant level, so any time the program audio is softer, the noise will be better isolated. If you can post some samples, I'll tell you what it is.