REW Questions

[Earfonia]

Most likely is that the Windows volume controls for input and/or output are set differently. Should also check that the View preference "Full scale sine rms is 0 dBFS" is set the same for all installations. Probably better using ASIO drivers to remove Windows audio settings as a source of differences.

Noted. Thanks John! I will check that.

 

[nhs]

Hi Folk,

I pull this thread out of over 2 years ago. I want to measure and create the sweep with REW.
Now, I would like to know if the DRC written by Denis Sbragion can work with the impulse response exported from REW?

Thanks

 

[ernestcarl]

Hi Folk,

I pull this thread out of over 2 years ago. I want to measure and create the sweep with REW.
Now, I would like to know if the DRC written by Denis Sbragion can work with the impulse response exported from REW?

Thanks

Yes. It can work… I’ve done it before, sort of. But I “cheated” and used DRC designer for the filters… didn’t particularly like the results in any combination and the additional 170 ms or so time delay so I stuck with simple manual mixed phase corrections instead.

 

[edechamps]

Now, I would like to know if the DRC written by Denis Sbragion can work with the impulse response exported from REW?

Yes, but IIRC Denis Sbragion's DRC is quite user-hostile when it comes to file formats. When I used it years ago it didn't even support WAV files, you had to convert the mono impulse response to "raw" bytes and be very careful to use the right sample type (32-bit float) and match the sample rate since raw files don't have any metadata to guard against mistakes.

DRC can be quite picky about text frequency response data (.frd) files as well, for example it would reject files that don't have points at 0 Hz and Fs/2, sometimes requiring manual intervention. This can be a problem for target curve files and especially microphone calibration files.

the additional 170 ms or so time delay

You should be able to reduce that delay by a factor of at least 10x if you make use of a more advanced convolver that uses a partitioned fast convolution algorithm. Back in the day I was able to run 32768-tap long DRC filters on 6 channels simultaneously with a delay <10 ms (combined processing delay + IR delay). That was many years ago though. I'd imagine modern desktop CPUs can do even better.

 

[ernestcarl]

I was able to run 32768-tap long DRC filters on 6 channels simultaneously with a delay <10 ms (combined processing delay + IR delay). That was many years ago though. I'd imagine modern desktop CPUs can do even better.

Interesting... I've heard of warped FIRs also... but, still, how are these supposed to work with such a minimal total processing delay if you want to for example, say, apply HPF below 20Hz without the phase changing and/or fix excess phase issues in the very low bass? I'm also left wondering why programs like Acourate and Audiolense do not seem to use something of the kind since the delay people usualy report is in the hundreds of milliseconds. Not that I've used those myself.

The way I've come up with to minimize the delay is by simply using less taps and minimum phase PEQs -- but it just so happens that my particular use case for "correction" is rather of the basic kind with no real need of "heavy-lifting" -- or so I think.

Couch Center MLP (FDW 15 cycles)

256 taps ~3ms delay and subtractive PEQs only for nearly all channels

Though, I am interested if it's possible to apply a linear phase HPF to the sub while keeping its group delay relatively untouched or as low as possible -- not that I need it. Just curiousity...

 

[dc655321]

You should be able to reduce that delay by a factor of at least 10x if you make use of a more advanced convolver that uses a partitioned fast convolution algorithm. Back in the day I was able to run 32768-tap long DRC filters on 6 channels simultaneously with a delay <10 ms (combined processing delay + IR delay). That was many years ago though. I'd imagine modern desktop CPUs can do even better.

AFAIK, convolution partitioning algorithms will reduce latency due to convolution itself, but they cannot help with the group delay present in linear-phase FIR filters. Or am I misunderstanding, and you were using min-phase FIR filters?

 

[edechamps]

Interesting... I've heard of warped FIRs also... but, still, how are these supposed to work with such a minimal total processing delay if you want to for example, say, apply HPF below 20Hz without the phase changing and/or fix excess phase issues in the very low bass?

AFAIK, convolution partitioning algorithms will reduce latency due to convolution itself, but they cannot help with the group delay present in linear-phase FIR filters. Or am I misunderstanding, and you were using min-phase FIR filters?

Sorry, I believe there's a misunderstanding - I was referring to the default options for DRC, which result in an IR delay of about ~3 ms. My understanding is that these defaults basically make DRC compute minimum phase at low frequencies and linear phase at higher frequencies. I did not mean to imply one can do linear phase at 20 Hz with no delay I was just referring to the delay induced by a less sophisticated convolver.

 

[nhs]

Thank you, all. That helps me a lot for first try.