DSP Robotics Support

Hi,

Wow that is cool !

How difficult would it be to change the code to allow for 5 channels to be recorded at the same time ?

6 or even 8 channels would work too (those are more like 5.1 and 7.1 surround sizes for example).

The wave file format allows for that see this link for details (I like this description):
http://web.archive.org/web/20080113195252/http://www.borg.com/~jglatt/tech/wave.htm

I would be willing to help if needed or if you want to keep it proprietary that's fine too, I may be able to pay for this as well depending on cost of course

I use Flowstone as a data acquisition system and need to record data between DC and 15KHz (16bit format @44100sps, or 12bit format minimum at 44100sps). I do it now by outputting the streams to Reaper and it works fine... however there are a few issues with the "ease" of using virtual cables, but that is another story

Thanks,

Aron

mccy wrote:Hey, this looks great! Thank you very much! Now I can handle the two arrays as dual mono mems/files, but how can I combine them to one stereo array?

thanks

Martin

Why do you want to do this?

aronb wrote:Hi,

Wow that is cool !

How difficult would it be to change the code to allow for 5 channels to be recorded at the same time ?

6 or even 8 channels would work too (those are more like 5.1 and 7.1 surround sizes for example).

The wave file format allows for that see this link for details (I like this description):
http://web.archive.org/web/20080113195252/http://www.borg.com/~jglatt/tech/wave.htm

I would be willing to help if needed or if you want to keep it proprietary that's fine too, I may be able to pay for this as well depending on cost of course

I use Flowstone as a data acquisition system and need to record data between DC and 15KHz (16bit format @44100sps, or 12bit format minimum at 44100sps). I do it now by outputting the streams to Reaper and it works fine... however there are a few issues with the "ease" of using virtual cables, but that is another story

Thanks,

Aron

Please PM me what you actually want to do, and I will try to help you...

I need stereo-sound, and it must be very simple, to save the recorded file as stereo and get it as a stereo array back in flowstone, because I want to switch between recorded and prerecorded material in the scratcher... I posted the "scratch" project in another thread.

1000 Thanks

Martin

Hey Youlean.

I'm glad that I found your module, however there are few problems to address.

Memory usage. Recording 15 minutes of audio (44.1k/stereo) produces about 600MB of memory usage. This is probably due to the green coversion on output (arrays of floats).

When recording stereo, channels are sent to output separately as arrays of floats. Converting this to single output file would require interweaving in greens (or whatever else in greens), which would produce additional usage.

Do you know any workaround for these issues?

The best way would be to have mono and stereo mem outputs, so that mem data are sent directly to saving or else, but from what I see - this dll prim does not supports such nodes.

So maybe there is some way through these red V (for ruby?) nodes and ruby?

Well, workaround for these issues is to build new wav recorder.

Actually it could be easier than I thought, and perhaps you can help with that.

Here is a scenario.

1. FS trigger activates recording
2. external dll records audio
3. FS trigger activates stop
4. external dll saves recorded audio to wav (some cache file with provided folder)
5. optional: external dll sends confirmation that file was saved
6. FS loads audio file from cache folder (standard guts for loading audio).

Everything else (cache cleaning and so on) can be handled from within FS.

So, the external dll:
1. receives trigger for recording
2. receives trigger for stoping (where stop = save file)
3. sends optional (i.e. this can be handled by delay in FS, but it's less safe) trigger that file has ben saved.
4. optionally - recording is set to mono or stereo.

Not too many interactions. No green arrays are created in FS. If on dll side there are no memory issues, then this should work pretty well, and should record "unlimited" audio.

Could you make such dll for audio recording?

You can't write directly to a file. You must write to a ram, but I can optimize this dll a lot, though functionality you have described above will require some time to implement, so I will be only available to do that for a fee.

Yes, I know, you write it either through some sort of buffers (faster) or put all to ram and from ram to disk at once (slower). Correct me if I'm wrong, and tell me whether this matters anyway.

But the point is, when you record 16bit audio (44.1k/stereo), then you get 600MB per hour. The problem I see right now, is - whatever is pushed through flowstone's greenery - produces a lot of ram usage, which is not really useful. Apart from that, some green prims don't release ram correctly. But with no audio inside (and with no prims that hold greenry data after unloading audio), this should work as a switch per switch (recorder/flowstone).

My questions.

From your perspective - how it will behave in terms of what I wrote above? (what should be I aware of?).

For how much and how long it will take?
(PM me, just point technical details in this post)

To other flowstoners:
who is interested to contribute, so that this could be then released on forum for all?
(PM me)

Can you first tell me what you want to achieve? Then I will tell you what will be the best approach using dll. PM me if it is sensitive...