Initial REW Measurements
After setting up Windows and installing the REW application, I test the hardware and software to ensure that REW functions correctly. The "fun?" part begins with measuring the room's audio performance, followed by an initial analysis to identify possible improvements. As a reminder, all the screenshots and descriptions were taken using REW Version 5.31.
Last Updated: 04/10/2025
Checking That REW's Working and Setting the Levels
​Assuming all the preferences have been set up (as described on this page), the next step is to ensure that REW can properly send test signals to the speakers and that REW can correctly receive audio from the UMIK-1. Once this is confirmed, the speaker volume is adjusted to ensure that the test signal is loud enough that the room's background noise does not distort the results, but not so loud that the speaker is overdriven.
​
REW's built-in SPL Meter and Generator perform these initial steps. These tools are invoked by clicking on the tool icon. They are shown at the top of Figure 1 and highlighted in yellow and red. Clicking on these brings up the dialog boxes shown below the toolbar in Figure 1. The Generator is highlighted in red on the left, and the SPL meter is in the yellow box on the right.
​​​
The Noise option is selected in the Generator dialog box, followed by Pink Random. The Full Range option is selected since the initial measurements will be full range (one speaker and subwoofer together). The level defaults to and should be kept at -12.00 dBFS. The desired speaker output should be selected near the bottom of the dialog box. In this case, the left front is selected. In this example, the left front is selected. When we're ready to check the levels (after setting up the SPL meter), the green play arrow is clicked.
​​
In the SPL Meter tool, the buttons below the numeric display should be set to SPL, C weighted, and S slow time averaging. If the UMIK-1 is appropriately connected, the number display will read some audio and likely change values as soon as the SPL Meter is opened. The meter reads the background noise, indicating that the microphone is connected correctly.
​
Hopefully, once the play button is clicked, the selected speaker will play pink noise. If not, the PC and AVR setup must be checked and fixed. With the pink noise playing, the SPL Meter's reading will increase. To set the proper volume level, the AVR's volume is adjusted until the SPL meter reading is roughly 75dB (see Figure 2).
​​​​​
Figure 1. REW with the Generator (left) and SPL (right) Dialog boxes opened.
Figure 2. SPL Meter Reading when Generator is Playing
Deciding What Measurements to Capture
​​​The general recommendation for capturing the room response with REW is to set up and take several measurements at different positions that cover the desired listening area. Specific recommendations vary between "experts" and depend on the use case (e.g., home theater, stereo listening, recording studio, etc.). For this initial measurement in my home theater, the goal was to gather data around the Main Listening Position (MLP). So, the plan was to take 5 or 6 data points at the positions shown in Figure 3. Data was taken for each of the front three speakers. (The location circled in green in Figure 3 was not used as it was too low in the seat.) A limitation of only taking data in a small area around the MLP is that other seats in the listening area are not measured and thus not considered. One can measure a larger area to rectify this and account for different seats. Measurements over a larger area could be done later.
​
When measuring a surround sound system, the predominant recommendation for the UMIK-1 orientation is to set it to point straight up and use the 90-degree calibration file. Using a boom microphone stand is almost a requirement because the boom stand makes it very easy to move a mic around the seating area. The procedure was to set up the mic at one position, measure the front left, right, and center speakers, and then move most to the next position.
​​​
Figure 3. P Multiple Measurements When Optimizing for MLP (Main Listening Position)
REW Measurement Tools and Graphs
​​​​Select the Measure icon from the menu bar (top left of Figure 4) to start. This brings up the Measurement dialog box (Figure 4, below menu bar), in which some settings should be updated before starting a measurement.
​
-
Type (yellow)—The measurement type is set to SPL.
-
Name (orange)—This field can optionally set a filename prefix, which helps track which measurement is which. The radio buttons to the right of this setting can automatically add some valuable information to the filename. The added fields are a number, date, and/or output indicator fields.​
-
Range (green)—The goal is to capture the home theater's frequency response, so the Range setting should be 20Hz to 20KHz. Figure 4 shows a slightly wider frequency range so that the subwoofer's rolloff below 20Hz can be seen.
-
Length Setting (blue)—This sets the length of the FFT. The larger the number, the more accurate the FFT and the longer the measurement takes. A setting of 256K is precise enough for this measurement.
-
Output (purple)—This setting determines which speaker is tested, and Figure 4 shows the right front speaker being tested.
​​
As for the remaining settings, the default values are acceptable for these measurements.
​​​
(Note: The computer has been set to dark mode, so the screen captures appear as they do. I like dark mode when it is supported, not sure why.)
​
Figure 4. REW Measurement Dialog Box with Key Paramters Highlighted.
A measurement results window opens after pressing the Start button and completing the measurements. An example is shown in Figure 5 (Note: This figure is after setting a 1/12 smoothing setting). When the window opens, the default view is an SPL graph. Before going into the details of the measurements and how they were taken, a few not-obvious tools on the window should be pointed out.
​
-
Along the top of the graph are buttons that allow you to select the type of chart you want to view. Figure 5 shows the SPL graph selected (yellow). Three other charts (circled in lavender) that are very useful for the initial measurements are the All SPL chart, RT60 chart, and the Spectrogram chart. Many also consider the Waterfall chart (blue circle) useful, but the same information can be (in my opinion) easier to see in the Spectrogram chart. Examples of these are described later.
-
Several controls (circled in orange and light blue) become visible when the mouse pointer hovers over the graph area. The +/- symbols zoom in (or out) the corresponding X or Y axis. The buttons at the bottom right of the graph change the X access between 10-200Hz or 20-20000Hz range.
-
In the upper right corner (circled in green) are a series of tools that can modify the graph's format. The default display format is reasonable, so modification may not be necessary. The Limits tool can help optimize and specify the chart's minimum/maximum display range.
-
Another important graph display setting is under the Graph menu (Figure 6). This menu can set the amount of smoothing applied to the graph. Without smoothing, the graph looks very "noisy," making it difficult to see the frequency response that human ears would perceive. Smoothing smooths the graph's peaks, making the average frequency response easier to interpret. The Smoothing options that (I think) are most useful are:
-
1/3 octave smoothing balances detail and readability. Suitable for general room analysis and EQ work.
-
1/6 octave smoothing is more detailed than 1/3 octave. It provides more details, helps identify specific problem areas, and shows more modal behavior
-
1/12 octave smoothing is suitable for a more detailed analysis of crossover regions and shows more detail in the bass region to aid in analyzing modal behavior.
-
Psychoacoustic Smoothing helps visualize frequency response in line with human perception, applying more smoothing at higher frequencies. It's useful for assessing how speakers will sound subjectively in a room.
-
​
Figure 5. REW Measurement Dialog Box with Key Paramters Highlighted. The graph uses 1/12 octave smoothing.
Figure 6. REW Graph Pulldown Menu Options
Taking Measurements
​Very simply . . .
-
Position microphone
-
Select the Measurement tool.
-
In the dialog box, confirm the settings, modify the filename, and select the output speaker. Then, click the Start button.
-
Repeat for each position, then return to #1 until all positions are measured.
​​​​​​​​
Reviewing the Measurements - Frequency
​​​​​
The first thing to look at is the SPL plots. ​The (messy) result of taking all 5 x 3 data points results in the SPL plot shown in Figure 7, which overlays all the data in one graph. This figure plots the results without smoothing and is not helpful. A few thoughts and observations on the SPL curves:
-
Not using smoothing can be helpful when focusing on frequencies below ~150Hz, but "smoothing" makes it easier to analyze the frequency response across the full frequency range.
-
Even with "smoothing" enabled, looking at the overlay of all data points is usually not very useful. It is better to look at the individual plots (previously discussed and shown in Figure 5) to see if some peaks or dips need fixing.
-
It is helpful to generate an average plot of all the data points. Any common peaks and dips in the individual plots that should be reviewed will show up on the average as a peak or dip. Figure 8 shows an example of the average frequency response graphs in Figure 7 (and added the 1/12 octave smoothing option). Some observations of this waveform are:
-
The Bass region between 25-100Hz is ~5dB louder. (This is probably because I boosted the bass after running Audyssey. 😃)
-
A dip at ~120Hz appears to be due to a room mode null in most measurements.
-
There are minor dips around 230Hz and 350Hz, possibly due to a room mode or Speaker Boundary Interference Response (SBIR).
-
The speakers are boosting frequencies above ~5KHz.
-
There is another common dip (between the datapoints) around 2.2KHz, perhaps due to the speaker's frequency response.
-
To address some variations in the desired frequency response, speaker and/or subwoofer placements, seating placement, or acoustic treatments need to be tweaked.
​​​​​
Figure 7. All the Measurements for the MLP (no smoothing)
Figure 8. Frequency Response Graph Showing Average of All the Measurements with 1/12 smoothing
Reviewing the Measurements - Time
​​Another essential facet of good home theater sound is understanding how the room reverberates. In other words, how long do sounds bounce around the room? Too much reverberation makes the audio sound muddled; too little makes the room's audio "dry" and unnatural. REW's RT60, spectrogram, and waterfall charts offer various methods for visualizing a room's time-domain response across its frequency spectrum.
​
Various reference materials recommend that the ideal RT60 value for a small room (the type of room being measured here) should be between 200-300ms, rising to ~500-600ms in the bass region. CEDIA's RP22 also contains a similar recommendation (see Chapter 9). REW's RT60 graph plots this information. In this graph, REW does not exactly create an RT60 graph because (as stated in its documentation) it is problematic to generate this data in the bass frequencies. REW generates a curve called Topt, along with other decay curves.
Figure 9 shows a typical plot for this room, and the Topt curve is highlighted. REW does not plot this curve based on the average generated for the SPL curve. This curve can be plotted for individual measurements. Overall, the results are very reasonable between 70Hz and 10KHz. The Topt value rises quickly below ~70-80Hz. While this result is generally OK, the decay times rise quickly, indicating a lack of bass trapping in the room.
​
Another way to interpret this information is through the Spectrogram chart, an example shown in Figure 10. This graph displays the energy at each frequency over time, with a color heat map representing energy levels (red indicates the highest energy, and dark blue represents the lowest). From this graph, one can deduce decay times at various frequencies. As can be seen, the decay times above 100Hz look to be below ~250ms, which is within the desired range. The chart reveals increases in decay times at lower frequencies, roughly 600+ms. Specifically, there are a couple of very long decay times around 25Hz and 45Hz. Even at the lower frequencies, the results are reasonable, although it would be helpful if those delay peaks could be reduced.
It is also possible to correlate the peaks and dips in the SPL graph with the corresponding frequencies in this chart. Notice that around 120Hz, there is a gap in energy, and there are smaller ones in the 200-300Hz range.
​​
Figure 9. P Average of All the Measurements with 1/12 smoothing
Figure 10. Example Spectrogram Graph
The last chart example for this page is a Waterfall chart (see Figure 11). This chart is another way to display the same information as in the Spectrogram, but in a three-dimensional surface model. The same observations mentioned above also apply here. The decay times are well behaved and less than 300ms above ~100Hz, and one can see a long decay around 25Hz, which may be partly due to a room mode.
​
Figure 11. Example Waterfall Chart.
Armed with measurement data for the front, left, and right front speakers, one could experiment with changes to improve the audio. Figure 12 shows the frequency response graph again, but this time, it labels some areas of the response that require further investigation. These areas could be addressed by changing speaker and subwoofer placement and/or adding acoustic treatments. Then, by rerunning Audyssey and retesting the speakers, the impact of the changes can be seen in the test results.
​
Further, one could use Audyssey's MultiEQ phone app to adjust the target curve or the MultiEQ-X Windows application to manipulate Audyssey's filters to compensate audio performance more directly. These are possible topics for future write-ups.
​
Figure 12. Frequency Response Graph Showing Average of All Measurements and with Comments on Aspects of the Response.