Contents - Index

Frequency Response 
Index

Use the OmniMic Frequency Response analyzer to measure the frequency response or impulse response ("IR") of a sound system.

Doing a Frequency Response measurement with OmniMic is easy -- essentially, you just play one of the proper Tracks, as indicated near the top of the form, set the microphone to pick up your system's sound, and the graph is shown on the screen.   The OmniMic software also allows many adjustments which you can use to customize the graph or measurement.  For more Advanced functions, see the help section on that topic.

Important! To measure Frequency Responses (or Harmonic Distortion, Reverb, Distortion, or BassDecay) properly you must be playing the specified audio file.  This can be a downloaded WAV or FLAC file from the Dayton Audio site, a stimulus WAV file you can save from the software itself, an OmniMic test track CD, or a test track DVD. 

Or the signal can be output by the OmniMic program itself through your computer's soundcard output or an attached DAC device. 
---> With OmniMic40k hardware, be sure to configure the playback soundcard on your computer to operate at 96ksps or higher!

For best operation, a high quality USB DAC, such as the Dayton Audio DAC01 can be used for this purpose. Beware of "audiophile" DACS, many of which fail to exhibit flat frequency responses!  48kHz sweep test files as used with original OmniMic, or a 48kHz DAC can also be used with an OmniMic40k, but proper results can only be obtained up to about 22kHz with those signal sources.

The needed audio file content is indicated at the top of the program window.   The measurement is synchronously matched to the test signals provided on the specified tracks, and will be unable to lock onto other test signals.  You can choose to use one of two types of signal by selecting between the "pseudo noise" and "sine sweep" buttons above the graphs.
  • Sine Sweep signals ("Short Sine Sweep") provide the cleanest and most accurate measurements, as well as being able to drive speakers at specific SPL levels.  This is the preferred choice for frequency response measurements and should be used for all high-frequency measurements
  • "Pseudo-noise" test signals, that sound like Pink Noise are easier on the ears for extended sessions.  The accuracy using pseudo-noise at highest frequencies, however, can be degraded by sample clock tolerances, so this type should only be used for rough measurements where sweeps may disturb others in the area.  Pseudo noise tracks must match the microphone type, 48kHz for original OmniMic microphones and 96kHz for OmniMic40k microphones.

    Frequency Response measurements will not operate correctly if you try to use other Pink Noise test signals or any signal other than the one specified on the program window.  The exact timing and spectral content of the signals are critical for proper operation of OmniMic's precise frequency response measurement system.

    About Frequency Responses
  • Frequency Response is a curve that shows how strongly an audio system reproduces different parts of the frequency range.  The curve will be higher on the graph at frequencies where the system plays louder, and lower (perhaps showing only varying background noise) at frequencies where the system plays weaker or not at all.
  • A perfect Frequency Response curve would look like a flat line over all frequencies.  A real-world one will have variations of some decibels ("dB") over most of the frequency range,  dropping off at very low bass or at high treble frequencies.  Ambient noise or dynamic range limitations may also cause irregular variation at the frequency extremes.
  • You can measure near individual speakers (about 1 meter away is best), or out in the room at various listening positions.  It is best to have only one speaker at a time playing to prevent interference and cancellations between audio waves.
  • The frequency response of a loudspeaker will be different at different places in the room.  You can also use OmniMic to make an ongoing "Average curve" over multiple positions (advanced) to give a typical frequency response curve.
  • A frequency response is related to its "impulse response" (the pressure signal that a system would make if it were fed by an extremely short pulse).  OmniMic calculates effective impulse responses mathematically by comparing a system's output with the signal it was stimulated with (either a swept tone or a noise-like tone).  OmniMic shows impulse responses, and you can select how much of an impulse response you want the analyzer to look at ("windowing") when comverting back to a frequency response.  Click on the impulse response (IR) graph (below the frequency response) at the latest part of the impulse response you want OmniMic to include when it computes frequency response. Shift-click on the impulse response to select where you want the IR analysis to begin. The selected portion will appear as a red trace, the rest will remain black.  The shape of the 'windowing' that tapers down the impulse response is shown in green or gray.
  • OmniMic can automatically set the edges of the windowing region, by checking the checkbox marked "aut". .  The automatic windowing will be turned off if you set the edges by clicking on the IR graph.
  • Reflections of sound in a room will appear as abrupt features in later parts of the impulse response (starting usually after about 3 to 10 milliseconds from the main impulse peak).  The response of the speaker without room reflections is smoother and less varied than when room reflections are included.  Removal of room reflection effects is possible only at higher frequencies -- at lower frequencies, typically below 200Hz, the reflections happen before even one full cycle of a longer low frequency wave has completed so they cannot be separated.
  • Because all audio frequencies are played at once in the pseudo noise signal, you cannot determine that the speaker is playing at any specific sound level at any individual frequency, so measurements with the pseudo noise show only relative response flatness.  Those curves are shown as values given in units of "dB".  Also, minute variations in clock frequencies of different players can cause the response at higher frequencies can appear to vary when using pseudo noise.  When high frequency accuracy is important, use the sine sweep signal.  (Pseudo noise stimulus is intended for use only in situations where the sweep tones would be too disturbing by others in the room).
  • To evaluate loudspeaker how frequency response changes at specific sound volumes and for best accuracy at high frequencies, use the sine sweep to display curve values as "dBSPL" (sound pressure level, as detected at the microphone).  With an OmniMic Photolink, you can easily set the drive level of the sweep to the required 2.83Vrms for true accurate sensitivity measurements.  There is also a "dB Sensitivity" button on the Frequency Response screen, which can be used to configure the display for speaker sensitivities when driven at the standard 2.83Vrms level.  The control this button brings up enables the program to compensate for measurement at different distances than the standard "1 meter".
    Frequency Response Options:
  • "all":  this setting shows a frequency response that includes all room echoes and reflections.   In other words, the entire impulse response.  In this view, impulse response itself is not shown.
  • "only to": for suppressing reflections. This is calculated only from the impulse response within the time selected.  To select a different ending time,  click the mouse at the desired time point within the impulse response graph. Impulse response information after the selected time will be excluded from the frequency response calculation.  Select this time to totally exclude later reflections visible on the impulse response plot.   Lower frequencies can't be measured using this option, limited by the length of the time selected.  This mode works best when the OmniMic is relatively close to a loudspeaker.
  • "blended": blends the "only to" impulse response with a time/frequency dependent filtered version.  In other words, this mode removes echoes when it can but gradually allows more of the IR time to be included at lower frequencies when it can't.  Energy toward the left side of the impulse response is given priority.
  • smoothing:  You can choose to smooth the frequency response graphs over 1/2 octave to 1/96th octave regions, to vary the amount of detail shown.  An unsmoothed frequency response will be very ragged except when echoes are windowed or the measurement is taken very close to the loudspeakers.  An additional "ERB" smoothing option simulates the sensitivity of human hearing to response irregularities for estimating the sonic importance of frequency response irregularities.
  • markers: When the frequency response display has the mouse's "attention", a blue text box will read out the frequency and level related to the location of the mouse cursor.  If you then type one of the keys 1 through 9, a persistent white "floating" marker will attach to the live (currently being measured) trace at the freqeuncy related to the marker.  Click again on the persistent marker to clear it, or type the 0 (zero) key to clear all of the persistent markers. 
  • Frequency Response measurement curves can be saved in FRD format using the "File, Save" menu.  You can save either or both of the current "live" curve (currently produced by the microphone) or any displayed Average curve.   These can then be reloaded using the "Added Curves" menu so that multiple Added Curves can be shown on the same graph along with the "live" curve and the Average curve.  When Frequency Response graphs that contain Added Curves are printed or saved in a "Snapshot", a list of legends can be included.  In snapshot files, a text note can also be added.  
  • Up to 40 saved "AddedCurves" Frequency Response curves (previously saved in "FRD" files) can be displayed on-screen simultaneously with the currently measured curve trace -- use the "AddedCurves -> Add" menu.
  • The full set of AddedCurve filenames, along with applied offsets or delays can be saved into a Curve List for easy retrieval to the OmniMic screen.
  • You can alter the thickness of the displayed "live" Frequency Response curve by toggling (left-clicking) the small button that is just to the right of the "OmniMic" logo at bottom left of the graph.  Right-clicking the button will change the thickness of the saved or average curves.
  • There are also a considerable number of additional features for advanced users available including phase responses, color polar displays, compression plots and waterfall decay plots.