Measuring Speaker Response in Listening Room

[texbychoice]

Have been working with REW and UMIK-1 mic for about a month. Think recent results are more accurate, but unsure. My listening room is roughly 10 x 12 with a sloped ceiling. Windows along one wall. Room treatment is needed, but not likely as it is multiple use room. Near, far, listening position all yield quite different results, so thinking due to the room. Recently did listening position test and went through the EQ process. Adjusted my Yamaha equalizer as close to the recommended EQ as possible. Sound was horrible, thin, no depth, and imaging non-existent. Adjusted the EQ by ear and back to great sound.

Read multiple tips, guides, suggestions for REW and UMIK-1 setup and measurement process. Is it even possible to get meaningful measurements in that room?

 

[John Mulcahy]

You should check whether your Yamaha EQ corresponds to the equaliser selection you were using in REW, definitions of Q vary between implementations. Make some loopback measurements of a few settings for a single filter on the Yamaha and see which of REW's EQ options produces the same filter shape for the same settings.

 

[Wayne A. Pflughaupt]

The room “is what it is,” so any room measurements are meaningful.

It’s not clear if you’re trying to adjust full-range or just a subwoofer. Either way, you should be concerned about primary listening positions. There are a number of approaches. Some prefer to concentrate on the main listening position. Some like to measure at all listening positions and get an average measurement. Some use a “moving mic” technique to accomplish a measurement.

If you’re equalizing full range, see my posts in this old thread from Home Theater Shack that offers some tips:

Spridle’s Experiment

One thing not mentioned in that thread, it may be best to use identical filters for the left and right speakers above ~300 Hz (or whatever the transition frequency is in your room). I’ve found that trying to EQ the two speakers separately in the upper frequencies messes with the imaging. That may be the reason you were unhappy with the imaging. Below the transition frequency, you should be able to EQ each speaker separately, if needed.

Regards,
Wayne

 

[texbychoice]

Thanks. Read Spridle's Experiment. Good information that makes me feel better about trusting my ears.

Backing up to some basics. Found conflicting advice about PC sound card calibration and microphone orientation.

When using a USB mic (UMIK-1), it is stated no soundcard calibration is required. It was stated PC soundcards are "flat" enough. However, found some recommendations saying soundcard calibration is necessary even when using USB mic. As an experiment, connected PC soundcard headphone output to mic input and ran REW calibration. The profile is far from flat. Does that indicate a PC soundcard problem?

Read recommendations for UMIK-1 orientation that included aimed at the speakers, aimed at the ceiling, and angled up 60 degrees. Tried all positions with different results. Is there a single "correct" orientation?

 

[texbychoice]

You should check whether your Yamaha EQ corresponds to the equaliser selection you were using in REW, definitions of Q vary between implementations. Make some loopback measurements of a few settings for a single filter on the Yamaha and see which of REW's EQ options produces the same filter shape for the same settings.

Thanks. My Yamaha capability does not have the resolution to match the REW EQ profile. Tried to match overall best I could. For example, the REW profile recommended -3db at 632Hz, then +6db at 801Hz. Tried a near to speaker measurement with a sound absorbing shield. Shield on three sides attached to speaker extending around the mic to reduce room reflection. That revealed about a 10dB increase from 3K on, which confirmed my hearing a forward and sometimes harsh high end. Made a speaker crossover correction to bring the high end down, much smoother response and sound.

Measurements in the listening position, full range with sub, showed some pretty big swings under 1KzHz. Probably the room?

 

[John Mulcahy]

Thanks. My Yamaha capability does not have the resolution to match the REW EQ profile. Tried to match overall best I could. For example, the REW profile recommended -3db at 632Hz, then +6db at 801Hz. Tried a near to speaker measurement with a sound absorbing shield. Shield on three sides attached to speaker extending around the mic to reduce room reflection. That revealed about a 10dB increase from 3K on, which confirmed my hearing a forward and sometimes harsh high end. Made a speaker crossover correction to bring the high end down, much smoother response and sound.

Specific EQ corrections are generally only useful below a couple of hundred Hz, above that any corrections should be restricted to tone shaping, which sounds like what you have done. For those kinds of corrections generally want measurements that minimise the room's contribution so the direct response of the speaker can be seen and adjusted if necessary.

Measurements in the listening position, full range with sub, showed some pretty big swings under 1KzHz. Probably the room?

Yes.

 

[DanDan]

Using a USB Mic there is no soundcard, no calibration, no loopback possible. The UMIK comes with Cal files for 0 and 90 degrees. Chose which way you want to point it and use that Cal File.
+10dB HF boost seems incredible.

 

[texbychoice]

Using a USB Mic there is no soundcard, no calibration, no loopback possible. The UMIK comes with Cal files for 0 and 90 degrees. Chose which way you want to point it and use that Cal File.
+10dB HF boost seems incredible.

Have tried the UMIK at 0 and 90 degrees using the correct calibration file. Forget where, but read a calibration file should be generated for PC/Laptop sound cards. Can generate one connecting headphone out to mic in. Far from flat. Definitely skews measurements. So, stopped using a soundcard calibration file.

Agree, 10dB seems unreasonable. Per crossover design software should have been about 2dB up from 3KHz. However, listening to various tracks high end did sound too forward and a little harsh at times. The measurement confirmed my ears, but do have a hard time believing 10dB.. Changed a resistor value in the crossover design software. Showed a gradual slope down from 3KHz. Made the physical change in the crossover. Measurement showed about 2dB up plateau from 3KHz to 16KHz where roll off began. Much smoother and balanced sound quality. .

REW has been a very helpful tool to improve sound quality in my listening room. However, am I correct that measuring only the speaker response requires an anechoic chamber or a quiet outdoor setting?

 

[DanDan]

Simply put yes. Outdoors up on a pole. If the speaker is powerful and there is little wind and or good protection, it doesn't really need to be super quiet. Sweep measurement 'see' only one frequency at a time so are very forgiving of chasing background noise. Yup a simple resistor is often all it takes. Speakers with Biamp terminals allow the resistor right there, no need to access the Xover.

 

[Sonnie]

Are these home/custom built speakers... curious as to why you would want a near-field measurement? Measuring the listening position is most common, and where EQ correction would normally be applied. It is very important to make sure the mic is in the same position before and after EQ to see what the EQ does. If you move the mic in between the before and after measurements, the corrections will likely be skewed.

Personally, from many systems I have heard, there is typically a 10-15dB drop from 20Hz out to 20kHz, with 20kHz being lower. A flat response, or boosted upper end would no doubt be harsh if I were listening to it, and my hearing is weak above 8-10kHz, depending on the ear. Mine is more like about 15-20dB drop, because I like more bass.

 

[DanDan]

+1 Gazillion Sonnie

 

[texbychoice]

Are these home/custom built speakers... curious as to why you would want a near-field measurement? Measuring the listening position is most common, and where EQ correction would normally be applied. It is very important to make sure the mic is in the same position before and after EQ to see what the EQ does. If you move the mic in between the before and after measurements, the corrections will likely be skewed.

Personally, from many systems I have heard, there is typically a 10-15dB drop from 20Hz out to 20kHz, with 20kHz being lower. A flat response, or boosted upper end would no doubt be harsh if I were listening to it, and my hearing is weak above 8-10kHz, depending on the ear. Mine is more like about 15-20dB drop, because I like more bass.

The speakers are in the vintage category, commercially produced. Due to age started with replacing electrolytic caps with film. Ended up rebuilding the crossovers with good quality caps, air core inductors, and good quality resistors. X-Sim software to generate response curves with component value changes. I took many measurements with REW and UMIK as a learning exercise.