Acoustics‎ > ‎

Measurement

Psycho-acoustics
The human ear, and brain, is capable of great sensitivity and ability to filter nuances from what we hear. This is a wonderful ability that enhances listening to music but it can be very deceptive when attempting to use hearing to measure acoustics. Our ears don't lie, but we often hear only what we want to hear or even hear what we thing is there but is not.

Measurements
Precision, measurement instruments are necessary tool which, when used correctly and knowledgeably, can give us a more UN-biased perspective.

RPlusD / ETF-5
This is a very useful application for measuring room acoustics from Acoustisoft; written by Doug Plumb. However, room acoustics measurement is not a completely intuitive process and the application can feel complex because of it's many features. So, here is a simplified view along with a set of directions which may help to more quickly get some useful information about a room.

The Concept
Fundamentally, ETF sends a complex signal of pink noise and a swept-frequency sine wave to one of your monitor speakers and collects the room response signal via a flat-response measurement microphone. Note a few details about this:
  • Note: Your monitor speaker is assumed to have an ideal, flat-frequency response. It doesn't ... but it is likely much better than the room's response -- so ETF ignores any speaker error. It is also very difficult to measure a speaker because you need a near ideal acoustic space and a precision, calibrated measurement microphone. This is another reason why good near-field monitors are well worth the investment.
  • All the various plots are derived from the same measurement data. So, after everything is set-up, take a measurement and save the data. That data can be recalled later for further analysis.
  • You don't need an expensive (>$1k) mic to get measurements of sufficient accuracy and resolution to find problems with room acoustics. The $50 price class Behringer ECM8000 does very well.
  • Generally, take 3-5 measurements which all look similar; avoid or reject the really 'odd' plots as noises can disrupt the measurement. As long as you can easily get a 3-5 series of plots that look really similar, the data is probably a valid representation of room acoustics -- at that spot.
  • Average 3-5 plots. Their differences are meaningless and just part of the noise.
  • Measurement position matters more before the room has been acoustically treated. So, it is best to just place the measurement mic at the "listening position" and that will give you easier comparisons.
  • An initial "loop-back" configuration with the sound card in your computer IS NECESSARY. This is how a reference is established for all additional measurements.
  • All measurements are the ratio between the Left Channel Reference (hard-wired loop-back) and the Right Channel Measurement (using the microphone).


The Sequence
The following sequence has been useful as a quick-reference guide for making ETF measurements. When it can be weeks between measurements (It takes time to construct new acoustic treatments like 8 bass traps!) it is easy to forget a step.

  • PC / Audio System Setup
    • Sound Card
      • SigmaTel
      • 48 kHz Sample Freq
    • Loop-back Test Config
      • PC LEFT OUT --> PC LEFT IN
      • PC RIGHT OUT --> PC RIGHT IN
    • Start : Programs : Accessories : Entertainment : Volume
      • PC Volume Control:
        • SPKR: 3.5 (1-7 scale) [REF]
        • WAVE: 3.5 [REF]
      • PC Recording Control
        • LINE: 3.0 (1-7 scale) [REF]
        • Mic: MUTE
      • Mic
        • Behringer ECM8000
        • 48V phantom power
      • PreAmp
        • Studio Projects VTB1
        • Input Gain (listening Position): 40, Speaker Test: 30
        • Blend: Full Left (none)
        • Output Level: 0 (mid)
  • ETF App Sequence
    • Initial Setup
      • Options : Mixer : Off
      • View : Frequency Response : Bode/FFT
      • Set display colors
        • White (Background)
        • Blk, DrkBlu, DrkGrn, Drk Vio, DrkOrg, Red, DrkGry, LtGry
      • Set display view range
        • single click = max zoom out
        • boxed area = zoom in
        • boxed area > graph size = zoom out
      • Config Loop-Back Test
        • PC LEFT OUT --> PC LEFT IN
        • PC RIGHT OUT --> PC RIGHT IN
      • Run Loop-Back Test
      • Verify the following levels (loop-back reference check)
        • Reference Levels:
          • -6 dB Mic Right (-6.1 meas)
          • >25 dB SNR Right (31.2 SNR meas)
          • -6 dB Ref Left (-6.4 meas)
          • >25 dB SNR Left (30.7 SNR meas)
          • 0 meters mic/spkr distance
      • Config for Room Measurement
        • PC LEFT OUT --> PC LEFT IN (Reference)
        • PC RIGHT OUT --> Right Pwr Amp --> Right Spkr
        • PC RIGHT IN <-- VTB-1 Mic preamp out
        • Position Mic at listening position
      • Config PC SoundCard Levels
        • PC Volume Control:
          • SPKR: 4.5
          • WAVE: 6.0
        • PC Recording Control
          • LINE: 3.0
      • Run Measurement
        • Verify the following levels (loop-back reference check)
        • Measured Levels:
          • -5 dB Mic Right (-1.2 meas)
          • >25 dB SNR Right (25.0 SNR meas)
          • 5 dB Ref Left (4.5 meas)
          • >25 dB SNR Left (30.9 SNR meas)
          • 2 (check that it's reasonable) meters mic/spkr distance(1.922)
          • Take 3 to 5 measurements
          • File : Save : Measurement Data
  • PC / Audio System Reset
    • Start : Programs : Accessories : Entertainment : Volume
      • PC Volume Control:
        • SPKR: 3.0
        • WAVE: 3.0
      • PC Recording Control
        • LINE: 3.0
  • Frequency Response Plot:
    • View : Freq Response : FractOct
    • Set display range as needed
  • Impulse Plot:
    • View : Time Response : Impulse
    • Set display range as needed
  • Waterfall Plot
    • View : Freq Response : Bode plot
    • Select Averaging
    • Set display range
      • 10 Hz - 400 Hz
      • +10 dB - -80 dB
    • Gate time:
      • 259 mS (long gate) (Set to 2x what you want to view)
    • View : Waterfall : ok to re-calc