An Engineer’s Comment on the High-Frequency Noise Spike Issue

The professional engineer in question has requested anonymity, but here’s what he had to say:

Studios are full of HF signals that alias back into the audio spectrum.
The guess about a TV monitor
somewhere in the mastering studio is valid indeed, as 15,750 * 4 -
44,100 = 18,900 Hz. It could even be a nearby TV transmitter.

Such “birdies” of constant frequency are relatively easy to trace, but
wideband HF is harder. For example, decent condenser mikes and
preamps go well beyond 100kHz at their output. Feeding them to
poorly-filtered ADCs (nearly all of them are!) can produce -40 to -60
dB of aliasing into the audio spectrum. As a side note, this is a
killer justification for ribbon mikes on “comb-spectrum”, high dynamic
range music such as authentic-instruments baroque.

Studio engineers can shrug and call HF birdies “digital sound”, yet to
me everything is analog: some within the audio band as intended, and
the rest just gets there unintentionally. Higher sample rate (192 or
176.4ksps for baroque, 88.2ksps for everything else) dramatically cuts
most of the latter.

See the comments of the previous post for more ideas.


Monday, February 13th, 2012 Audiophile, Bob, General, Technical

2 Comments to An Engineer’s Comment on the High-Frequency Noise Spike Issue

  1. Hey Bob,

    If the spikes in any one song are constant amplitude they may very well be “redithering,” especially if they of a certain age (as is Mr. Stingy). UV-22 was very popular for about 10 minutes and, although it dumped a boatload of energy at 22 kHz, there were and are other “shaped” redithering schemes that concentrate all their energy at one very high frequency.

    Ditheringly yours,

  2. Oliver Masciarotte on February 14th, 2012
  3. Thanks, Oliver.

  4. Bob on February 14th, 2012