Taking and Interpreting Measurements with REW (FREE eBook)

[Keith_W]

BTW guys I didn't know where to chuck this, so i'll put it here.

The book contains a formula to calculate the Schroder frequency. But it's easier to ask Copilot! That is the correct formula, and that is the correct result.

 

[tw99]

@Keith_W I've just read through the shorter eBook - very useful indeed. One question I have is about the SPS measurement that you recommend doing even if using MMM.

I originally measured my speakers individually using MMM and created filters directly from that MMM output - in this context, how would I use the SPS measurement ?

 

[Keith_W]

@Keith_W I've just read through the shorter eBook - very useful indeed. One question I have is about the SPS measurement that you recommend doing even if using MMM.

I originally measured my speakers individually using MMM and created filters directly from that MMM output - in this context, how would I use the SPS measurement ?

The MMM does not capture any timing information. If you want to make any timing corrections, e.g. time align a subwoofer, make phase corrections, etc. you will need a SPS. Also, no timing information also means no decay measurements - can't examine impulse or step response, waterfall, spectrogram, RT60, etc.

If you want to make DSP corrections, the correct way is to take some kind of spatially averaged measurement - either an MMM or multiple sweeps and average them - and use that to correct the frequency response. Which is what you did. But correction of the freq response is only half the story.

 

[janbth]

BTW guys I didn't know where to chuck this, so i'll put it here.

View attachment 504405

The book contains a formula to calculate the Schroder frequency. But it's easier to ask Copilot! That is the correct formula, and that is the correct result.

A comment about the Schroeder frequency: The formula given by Copilot is also given by Floyd Toole in Sound Reproduction. In the 4th Edition it’s in Sec. 5.1. He also writes: «The multiplier constant changed from the original 4000 to 2000 in the 1996 paper, emphasizing that this is not a precision calculation». My conclusion is to be careful with the Schroeder frequency, and be careful with Copilot, ChatGPT, etc.

 

[john61ct]

Microphone question, LMK if there is a better thread for them

Given multiple calibrated mic's plus timing loopback going through the same interface, are there / what are the / advantages of Multi-input capture from a practical POV?

Yes paying REW the $100 Pro license

Just speed / convenience automation, or increased accuracy, or does it open up new functionality?

Given a USB UMIK-2, can that also be used at the same time as those going through the interface?

Are there stages / methods where only one mic should be used?

How about when using Acourate?

 

[nanook]

There surely are experts around that are more experienced than me. Please add / correct me.

Given multiple calibrated mic's plus timing loopback going through the same interface, are there / what are the / advantages of Multi-input capture from a practical POV?

I only see the advantage to save time. Disadvantage could be reflections due to several mics + holders.

Given a USB UMIK-2, can that also be used at the same time as those going through the interface?

Be careful regarding latencies when using different AD-paths in parallel.
ASIO is quite reliable when using the same Audio-Interface and using timing-loopback on one channel assures you get the exact time-of-flight.
Using e.g. an audio-interface with Mic-Inputs and a Mic with integrated AD-conversion like UMIK-x, I'd expect different latencies almost for sure- maybe not even consistent between runs.

Just speed / convenience automation, or increased accuracy, or does it open up new functionality?

When using REW, you average frequency responses obtained either from sweeps or spectra taken with pink noise. Whether you do this averaging from parallel or sequentially acquired data should not make a difference. --> Just speed imho.

 

[Keith_W]

Given multiple calibrated mic's plus timing loopback going through the same interface, are there / what are the / advantages of Multi-input capture from a practical POV?

The easiest way to say this is: if you don't know why you need multi-input capture, then you don't need multi-input capture.

One example why you may need multi-input capture: the two microphone subtractive method for measuring subwoofers. The problem with measuring subwoofers is that the wavelengths are so long, and typical rooms are so small, that the reflection has already arrived back at the mic before the subwoofer has finished producing the test tone. To illustrate, the period of a 20Hz sine wave is 50ms. Sound can travel 17.2m in that time. For this reason, there are many schemes to measure subwoofers. One scheme involves using two microphones, spaced a known distance apart. The first microphone captures the direct sound from the sub, and the second captures the direct sound with a slight delay. You then use the two measurements to subtract room reflections. This is how the Audio Chiemgau Mode Compensator works.

Another reason would be to capture reflections. A single mic will capture the reflection, but you don't know where it came from. If you had two or more mics, you can tell. This is why the Trinnov microphone looks so strange.

Are there stages / methods where only one mic should be used?

99% of the time a single mic is sufficient, if not 100% of the time for most hobbyists.

How about when using Acourate?

Acourate requires an XLR microphone, so you will need an interface. Note that DSP lives or dies with the quality of your measurement. Not everyone agrees, but I think that you should ONLY use individually calibrated microphones for DSP. It's less critical for room measurements.

 

[john61ct]

Acourate requires an XLR microphone, so you will need an interface.

Yes I thought that was implied here

> Given multiple calibrated mic's plus timing loopback going through the same interface

Sorry if that was not clear enough

 

[john61ct]

Fair enough, thanks for taking the time.

The easiest way to say this is: if you don't know why you need multi-input capture, then you don't need multi-input capture.

 

[jeffaegrim]

@Keith_W, thank you so much for writing this - it is the go-to starter resource for interpreting REW measurements and I have sent the book to a bunch of people already. Have you considered writing more on the topics briefly covered in the addendum? There are lots of grey areas, and your thoughts as a DSP aficionado can only be parsed through forum posts. Which is perfectly fine, now that I think about it, considering they usually provide so much value for free

 

[Keith_W]

@Keith_W, thank you so much for writing this - it is the go-to starter resource for interpreting REW measurements. Have you considering writing more on the topics briefly covered in the addendum? There are lots of grey areas, and your thoughts as a DSP aficionado can only be parsed through forum posts. Which is perfectly fine, now that I think about it, considering they usually provide so much value for free

You're welcome. The original intention of the book was to expand on Amir's thread Room measurement tutorial for dummies, where Amir left us hanging after writing Part 1 many years ago There were many requests in that thread for Part 2 and more. At first I was going to write a long forum post, then I realized there was too much to cover. Then it became a long article, and it still wasn't enough. Then it became a book, and it STILL isn't enough. I had to stop somewhere, and I decided to stop ... at 83 pages. If you want to learn more, you'll have to pay! You need an acoustics textbook like Everest, or go and buy Toole. Some of the information provided in my book isn't covered by either of those books. But my eBook doesn't go in as much depth.

 

[jeffaegrim]

You're welcome. The original intention of the book was to expand on Amir's thread Room measurement tutorial for dummies, where Amir left us hanging after writing Part 1 many years ago There were many requests in that thread for Part 2 and more. At first I was going to write a long forum post, then I realized there was too much to cover. Then it became a long article, and it still wasn't enough. Then it became a book, and it STILL isn't enough. I had to stop somewhere, and I decided to stop ... at 83 pages. If you want to learn more, you'll have to pay! You need an acoustics textbook like Everest, or go and buy Toole. Some of the information provided in my book isn't covered by either of those books. But my eBook doesn't go in as much depth.

I was thinking about getting the new edition of Toole as a hardcover, but the prices are insane. It will probably have to be a softcover. And, from what I remember reading (which was perhaps ameliorated in the fourth edition), advanced DSP for domestic use wasn't a topic broadly covered

 

[Keith_W]

I was thinking about getting the new edition of Toole as a hardcover, but the prices are insane. It will probably have to be a softcover. And, from what I remember reading (which was perhaps ameliorated in the fourth edition), advanced DSP for domestic use wasn't a topic broadly covered

Unfortunately there are no good books for advanced DSP for domestic use on the market. The only one I can think of is Mitch Barnett's Accurate Sound with DSP, but that book is flawed. It's too focused on Acourate, does not attempt to explain the difference between linear phase and minimum phase, or how to take measurements. A better book is needed. I wrote one, but I don't have the confidence to publish it. It is also so bloody long that it's unwieldy. So I have been releasing snippets of it for free.

 

[respice finem]

Unfortunately there are no good books for advanced DSP for domestic use on the market. The only one I can think of is Mitch Barnett's Accurate Sound with DSP, but that book is flawed. It's too focused on Acourate, does not attempt to explain the difference between linear phase and minimum phase, or how to take measurements. A better book is needed. I wrote one, but I don't have the confidence to publish it. It is also so bloody long that it's unwieldy. So I have been releasing snippets of it for free.

Just s thought: Maybe it would make sense to issue the full book for the more technically savvy, and an AI "condensed" searchable PDF/ODF version as a free download for the rest?

 

[MaxwellsEq]

A better book is needed. I wrote one, but I don't have the confidence to publish it.

Perhaps you and @amirm should publish some ASR books?

 

[jeffaegrim]

Unfortunately there are no good books for advanced DSP for domestic use on the market. The only one I can think of is Mitch Barnett's Accurate Sound with DSP, but that book is flawed. It's too focused on Acourate, does not attempt to explain the difference between linear phase and minimum phase, or how to take measurements. A better book is needed. I wrote one, but I don't have the confidence to publish it. It is also so bloody long that it's unwieldy. So I have been releasing snippets of it for free.

Keith, I completely understand your commitment to academic publishing integrity, but I am equally intrigued by the contents of this book and believe that your collected effort should not go to waste - particularly during this historic moment of rising relativism and AI plagiarism, which already undermine the integrity of scientific inquiry - and owing to the aforementioned circumstance of literature on this topic being very sparse. This is why books have editions And perhaps you recall Wittgenstein, as an older man, walking back and disavowing the arguments made in his magnum opus, the tractatus logico philosophicus - which he wrote when he was younger - simply because he changed his mind. No shame in that either.

 

[john61ct]

Hear hear.

Even if just "by donation" buy me coffee

And the title

"My opinions on advanced DSP for domestic use - I could be wrong!"

 

[Keith_W]

I will release the book when I am confident it does not contain any mistakes. I am not in the business of misleading people.

Anyway, I don't know where else to put this, so i'll put it here. Over the weekend I was invited to measure a friend's system and comment on what can be done to improve it. The system has the whole shebang: audiophile power cords, cable elevators, tube amps with very expensive tubes, and FOUR SETS of speaker cable, each set costing more than $1k. Why four sets? Because they sound different and he picks the set that he is in the mood for that evening.

So I sat down to listen. He saw me pull a face and asked me what's wrong. I said the soundstage sounded really weird and the tone was all off. I then pulled out the measurement setup and plugged it all in and did a sweep in REW. I was going through the tabs when I saw this:

The polarity of the left speaker was inverted compared to the right. I asked him to check the cable connection. Red to red, black to black - all checked out. I made him check every cable in the system. So I wondered whether he had a cable which was incorrectly manufactured. So I got him to remove every cable in his system and I checked each one with a multimeter. All his cables worked just fine. This means that the manufacturing defect which caused one channel to have inverted polarity was going to be in his electronic components or his speakers!

So I asked him to reverse the polarity of the left speaker. This was the result. We sat down to listen. No need to guess at how much improvement there was. You should have seen his face when he realized that he had been listening to his system with the polarity inverted, it was priceless! I usually keep a professional poker face but I couldn't hide the absurdity of it all. This will be one guy who won't be telling me that the reason I can't hear the difference between cables is because my hearing isn't as good as his He asked me which is the faulty component? Preamp? Power amp? Speakers? I told him I don't know, I would have to test each one, and it would take me hours. And I had already spent a long time testing all his cables.

He shall remain un-named because I don't want to cause embarrassment.