After hearing someone mention how most people only look at the frequency response graph in REW when working on treating their room or setting up speakers, I figured that it might be helpful for me to run through how I use REW for my own studio and when I’m consulting for others. What I won’t be covering here is how to deal with the problems that you might see in the graphs as well as what are acceptable results.
At the most basic level I look at the frequency response graphs to get a general idea of how a system is behaving. In this case I’m looking at the overall shape of the curve, any large bumps or dips, and where the system rolls off. I find this graph useful for finding dips that may need to be treated. This is mostly a starting point but it’s also useful when moving speakers around the room and trying to find the best spot. When looking at this graph it’s important to set the scale of the Y axis. If the range is too large, anything will look like a flat line. This is one way which many people try to deceive others. The smoothing is also important to set. If it’s set too high then again, anything will look like a flat line. I generally use 1/24 smoothing keeping in mind that many of the fluctuations shown with this smoothing aren’t audible. The psychoacoustic smoothing gives a better idea of what’s audible.
The graph which I find to be the most useful is the spectrogram. This is where I look at decay times to see where there are potential problems. I generally avoid the waterfall graph. Although it can give a lot of insight into the decay behavior, the range and time window set will drastically affect how the graph looks so it can be finicky to set up. Again this is one way which people mislead others to make their room seem like it has no issues. I find the spectrogram to be more straightforward with the stock settings.
The decay graph is again another representation of the decay behavior which can be useful in seeing how the frequency response changes through time. Ideally each slice interval would have the same frequency response. The more that a slice interval deviates from the initial response, the more likely that there are decay problems.
The phase and group delay graphs can be very insightful but for the purpose of treating rooms, it doesn’t normally offer too much useful information that will affect treatment or speaker placement. I use these graphs primarily when designing speakers or setting up more advanced DSP correction systems.
The impulse graph is the most useful graph for understanding higher frequency reflections happening around the room. Both the dBFS and %FS display modes will make problem reflections visible and easy to see if treatment in those spots are effective. The timing of those reflections can be used to track down where in the room those reflections are happening.
The last graph that I frequently look at is the distortion graph. When treating rooms it doesn’t offer much useful information. This graph is mostly used for designing speakers or making sure that a system isn’t reaching its SPL limits. I mainly use it to compare speaker systems. It’s somewhat disturbing to see how high the distortion is on many high end speakers.
The RT60 graph with the EDT curve is one what’s frequently talked about and while it does provide some useful information, I find that it doesn’t give me anything that I’m not already getting from the other graphs.
When it comes to integrating subs into systems which I typically help studios with, I’ll be looking at the frequency response, phase, and time alignment using the impulse graph. These are also critical for setting up surround systems. It’s not always possible to get clean results given the constraints of the equipment being used for the crossovers and other DSP but there’s usually some room to play around with and find the best result. I believe I’ve mentioned in a previous blog post that simply adding on a sub to a system or going through the crossover built in to subs will usually cause more harm than good. It must be designed as a single cohesive system taking into account the natural rolloff of the speakers, the phase around the crossover, and time alignment of the speakers. In general the lower the crossover, the less critical these factors considering the larger wavelengths.
Hopefully this post can help studio owners get a better sense of what’s going on in their room instead of just looking at how smooth the frequency response looks. Of course at the end of the day, every system needs to be auditioned since our hearing is much more complex than what these graphs show. It’s one of the drawbacks of consulting remotely but using REW to make effective decisions and recommendations leads to a great starting point which can then be tweaked.