Hello Julien, On Thu, December 4, 2008 22:28, Julien Claassen wrote: > A short question: My setup is a simple microphone (live), to nearfield > monitors and a text-based Linux on a 1.8gHz CPU. Is it feasable for me to use > DRC? ...
with BruteFIR it is even overkill, unless you need really low convolution latency. In this situation consider Jconv, which uses a differenta algorithm and provides lower latency without increasing the CPU usage too much. My current convolver is an obsolete Celeron 800 MHz, which is slow even when compared to other Celeron CPUs of the same class because it uses some old notebook chipset on the MB. With BruteFIR it runs six 44.1 KHz 65536 taps filters consuming about 40% of the CPU power. With a simple stereo filter set the CPU usage was around 15%. The latency is set at 8192 samples though, it may be unacceptable for certain situations. Jconv should be able to achieve latencies down to the soundcard minimum acceptable value with little higher CPU usage, thanks to its non uniform partitioning algorithm The key issue is the mic. You need an omnidirectional microphone with a flat response to get accurate results. The cheapest option is the Behringer ECM8000, which is already enough to get pretty good results. There are other cheap calibrated mics on the market (IBF Akustik for example) providing even better results, optimal for most common situations, for just few more bucks. A true measurement microphone (B&K, ACO, EAW and so on) costing on the thousands is most of the times an overkill for this task. As usual, a good starting point is this one: http://www.duffroomcorrection.com/wiki/Main_Page Bye, -- Denis Sbragion InfoTecna Tel: +39 0362 805396, Fax: +39 0362 805404 URL: http://www.infotecna.it _______________________________________________ Linux-audio-dev mailing list [email protected] http://lists.linuxaudio.org/mailman/listinfo/linux-audio-dev
