Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[paul1970]

I was all pleased with myself for generating a WAV IR file (with 6 channels, no less) from REW (not an easy process first time through). After successfully enjoying the results with Redbook rips I now realise that 'things aren't right' when playing anything that isn't sampled at 44.1khz (the rate I'm working in in REW). After a bit of research I find that MC is capable of switching between IR files at different sample rates to match the audio being played, but it seems that REW can only operate at either 44.1 or 48k. I cannot justify the expense of Acourate or Audiolense no matter how terrific I'm assured they are.
Do I have any other options? Have I misunderstood the way any of this works?
Thanks.

[mwillems]

I was all pleased with myself for generating a WAV IR file (with 6 channels, no less) from REW (not an easy process first time through). After successfully enjoying the results with Redbook rips I now realise that 'things aren't right' when playing anything that isn't sampled at 44.1khz (the rate I'm working in in REW). After a bit of research I find that MC is capable of switching between IR files at different sample rates to match the audio being played, but it seems that REW can only operate at either 44.1 or 48k. I cannot justify the expense of Acourate or Audiolense no matter how terrific I'm assured they are.
Do I have any other options? Have I misunderstood the way any of this works?
Thanks.

Your convolution filters need to match the sampling rate at which playback is happening or you'll get wonky results (as you discovered), which requires multiple sets of filters.  It sounds like you can generate 44.1 and 48 in REW, but need something for the higher sample rates?  I know of two non-acourate/audiolense workarounds:

1)  Resample everything to 44.1 or 48 Khz in JRiver's Output Format DSP module.  This works for some people and not for others, and obviously results in not bit-perfect output.  When I enable resampling my convolution starts acting up (weird latency, etc.) so I don't use this approach, but some users have reported success with this method.

2)  If you don't mind fiddling, there's a free convolution filter generating tool called "RePhase" that can be used to replicate many of the same effects that REW replicates and can encode the WAVs at an arbitrary sample rate.  The catch?  It's not automatic or measurement based, so you'd need to manually dial in the filters that REW is applying which might be easy or kind of hard depending on the number and complexity of the filters.  I use RePhase to do linear phase crossovers without doing general room correction, and it's filters work a treat but it's definitely not as "automatic" as REW.

[Matt]

The JRiver convolution engine can resample your filter to any rate.

So in theory using the highest sample rate filter possible and letting it resample down when necessary should yield reasonable results.

[mwillems]

The JRiver convolution engine can resample your filter to any rate.

So in theory using the highest sample rate filter possible and letting it resample down when necessary should yield reasonable results.

Matt, do I understand you correctly that the convolution module itself can resample filters (i.e. it doesn't just switch between pre-made filters)?  I thought multiple filters had to be made.

Or are you referring to the resampling available in the output format module?

[Matt]

Matt, do I understand you correctly that the convolution module itself can resample filters (i.e. it doesn't just switch between pre-made filters)?  I thought multiple filters had to be made.

Both.

It searches for the best match if there are multiple filters, then resamples if the best match still isn't a perfect match.

[mwillems]

Both.

It searches for the best match if there are multiple filters, then resamples if the best match still isn't a perfect match.

Thanks for the clarification. That's a really neat feature that I didn't realize was there.  I'd always assumed it wasn't resampling because I got odd results trying to play 48 KHz material with a 44.1KHz filter (for example).  Any insight as to why playing at a sample rate other than the one the filters are made for might sound like "things aren't right" to OP (and me too for that matter)?  Is there some trick to it?

[paul1970]

So if I understand correctly, Matt is suggesting that re-sampling to a lower rate does not cause the problems that re-sampling to a higher rate does. So if you have a filter sampled at 192k there's limited value in generating a number of filters. I think that although JRiver will re-sample the filter to a higher rate the results of doing this aren't so desirable.

So my problem is that REW won't generate a filter >48k. Thanks, mwillems for your original response. I'll investigate RePhase as I'm confident I'm generating good filters in REW. If it's just a case of getting them into a format RePhase can use I'm inclined to write a small script/tool to do it. Shouldn't be rocket-science. In the meantime I'll try option 1 as a stop-gap and see how I get on. If that doesn't work I'll use the PEQ to get rid of the worst bumps until I can get to grips with RePhase. Now I've heard my setup without them I cannot live with them any more.

Matt, thanks also for your input. It seems that JRiver uses a file with a 'cfg' extension to point to a number of filters that can be switched. If I do want to create multiple filters then where can I found out what format JRiver expects that file to be? I presume it's a format that Acourate or Audiolense produces natively.

[mojave]

So if I understand correctly, Matt is suggesting that re-sampling to a lower rate does not cause the problems that re-sampling to a higher rate does. So if you have a filter sampled at 192k there's limited value in generating a number of filters. I think that although JRiver will re-sample the filter to a higher rate the results of doing this aren't so desirable.

I use Audiolense and take measurements at 48 kHz. Audiolense resamples the filter to other sample rates and saves the filter including upsampling to 192 kHz. After Matt posted yesterday, I created a filter at 48 kHz in Audiolense and then let JRiver resample the filter to 44 kHz. It worked perfectly. Whether you use Audiolense or Acourate to resample filters or use JRiver it really shouldn't matter much. I don't think you need to take a measurement at every sample rate and then create multiple filters.

If I do want to create multiple filters then where can I found out what format JRiver expects that file to be? I presume it's a format that Acourate or Audiolense produces natively.

Here is the format JRiver expects:

xxxx2.0_441
xxxx5.1_48
etc.

The regular expression is:
^(.+)(\\d{1}.\\d{1})_(\\d{2,3}).cfg$

Which outputs:
Name
Channels
Sample rate

So for a 2.0 filter you might call it "80 Hz XO Flat Target 2.0_441"