Hi, jpo234@xxxxxxxxxxxx wrote: > Patrick, > thanks for your quick reply. I'm not quite sure whether your answer > is really the answer on what I wanted to ask, so I'll clarify below. > > Patrick Shirkey <pshirkey@xxxxxxxxxxxxxxxxx> wrote: > >>Your be either one rme hammerfall + digital mixing desk with adat >>connectors. Otherwise you can use multiple cards but they need to be >>wordclock synced so you tracks stay in sample sync with each other. In >>that case you could use two rme hdsp multifaces, 2 midiman delta1010s >>and there are a couple of other 8 i/o cards with ALSA support. > > > I think to make sure that this is really what we need I'll have to > provide a better description of the problem. We are developing a > testbed for wireless voice transmission. We will have up to 16 active > phones, that are independently used to receive voice samples. Having implemented a tool very similar to this I'd say Patricks suggestions are pretty much correct. Back then the only viable alternative was RME Hammerfall 9652. I think it stands as a good alternative today also. Our solution was built to handle 72 channels (three hammerfall cards) and did it's analysis in realtime. As it happens we never went above 24 channels (one card), but that worked rather well. This was before jack so it was interfaced directly to alsa. If we where to redo it today I think we would have used Jack however. > We have > to record each voice transmission to a .WAV file (because this is what > the quality measurement tool understands). > To do this we will have to connect the 16 phones to a Linux box (prefered) > and need the ability to save the voice sample to a file. Each sample is > completely independent of all the other samples. Basically we need the > ability to do "sox /dev/phonespeakerX phoneXsample.wav" where 0 <= X <= 15. Since it seems you are mainly interested in creating WAV files, there may be a possibility to create a solution with of_the_shelf tools. I'm mainly thinking of ecasound. I'm not sure if it can handle the throughput but I think it would. Quite easy to find out anyway ;) Regards, Robert > > >>>Will we have 16 /dev/dsp devices [e.g. /dev/dsp0 .. /dev/dsp16]? >>> >> >>If you use jack you will have an easy to understand interface for >>routing the i/o's. > > > Following your suggestion I glanced at jack. I'm not quite sure whether > this is really what we are looking for. > > Thanks in advance > Joerg >
