Is a Room correction trial possible?

[brianuk]

Hi,

I'd like to 'trial' room correction and wondered if it is possible to be able to:

Record a room with laptop / sweeps / impulse.
Produce some configurations
Rip a CD
Process the ripped data with the results of the sweeps (ie the 'correction')
Burn new version of the CD

Simply, to trial DRC without having to build a hardware setup and to use my known CD player etc (like for like comparison).

Any pointers to software?

Thanks.

[ackcheng]

This software will do the job

http://www.acourate.com/

You can do a logsweep recording and send Uli the file, he will give you the corrected CD to listern

[lazycatken]

Originally Posted by brianuk

Simply, to trial DRC without having to build a hardware setup and to use my known CD player etc (like for like comparison).

Any pointers to software?

Sure, just burn a corrected CD for your CD player.

Per software, you may try
1. Generate log sweep test tone by glsweep included in DRC package.
2. Play test tone & record response of your system by Audacity. Put test tone on both channels. Loopback a channel as reference. Record each channel individually.
3. Generate impulse response by lsconv included in DRC package.

You can do above steps by HolmImpulse.

4. Generate filters by drc. You had better check weather drc gets the correct impulse center or not.
5. Do convolution by brutefir or convolver. You have to set appropriate attenuation to avoid clipping. Do normalization by sox then.
6. Burn CD and play.

To avoid clipping, all the steps, except burning CD, should work on 32 bits float format.

[brianuk]

Thanks very much. I might be back to ask for more help when it goes wrong

[ChrisA]

Originally Posted by lazycatken

Sure, just burn a corrected CD for your CD player.

THANKS. I had never thought of doing this. I understand all the steps required

I've have worked on software for a couple synthetic aperture radar projects at work. The algorithm is much the same. PC hardware is fast enough that you can do a frequency domain convolution in real time. There would be a "lag" for for music listening we don't care much about that. People are routinely running SDR (software defined radio) with 24-bit 96K sampled audio in real time on low end PCs This job is easier than SDR.

Recently I've been thinking about what to build and I had decided on a small single ended tube amp with a design that runs negative feedback all the way from the speaker output terminal to the input (global NFB)

But then I got to thinking that there had to be a way to make the feedback loop even longer and I've been thinking about using a microphone in the listening room to drive the NFB loop. But I think I'd never be able to stop it from squealing. It would never work. (Some frequencies would feedback positively.)

Ah, but this software system could work. Likely you'd have to place the mic within inches of the listening location and make sure the mic had the same directional pickup pattern as your ears but it could work.

[Key]

Originally Posted by ChrisA

Recently I've been thinking about what to build and I had decided on a small single ended tube amp with a design that runs negative feedback all the way from the speaker output terminal to the input (global NFB)

But then I got to thinking that there had to be a way to make the feedback loop even longer and I've been thinking about using a microphone in the listening room to drive the NFB loop. But I think I'd never be able to stop it from squealing. It would never work. (Some frequencies would feedback positively.)

Well you probably shouldn't actively monitor the mic(s) You can capture an impulse response and use that on the fly with the mics off. Or if you are set to use the mics come up with a side chain. Personally I think this is the wrong approach to the problem because your brain might already have a feed forward error correction built in.

I do like the idea of the loop from the speakers though.

[brianuk]

I believe the B&O speaker monitor themselves, but in real time I guess its likely to just be a lower set of frequencies?

Anyway my idea is to just burn a CD to try this as proof of concept before spending on a good looking PC box.

[Key]

Another problem with real time and digital I forgot to mention. There isn't such a thing as "real time" when it comes to digital. There will always be a small amount of measurable latency which would make it impossible for the mic/correction to act fast enough - at least I think so. So an impulse response would probably still be the best option for the areas that one would work - phase correction, and offsetting of certain errors. I don't think this works well though for room correction since that is time based and not static frequency or phase offsets. Anyway just pointing out some problems I see with your idea Chris.