Re: [AUDITORY] Online rhythm production experiments : follow-up

[Bob Masta <audio@xxxxxxxxxxx>]

Are you asking about an *online* system, or is this
something else?  If the latter, my Daqarta for Windows
package includes a macro app called Auto_Recorder that
may be of interest, even if only as a starting point.  The
Help topic is at <https://www.daqarta.com/dw_0o0a.htm>.
It includes the full source code, in Daqarta's macro
language.

Auto_Recorder uses a standard Windows sound card to
record 1 or 2 channels of 16-bit sound.  It is essentially a
"sound-activated recorder", with fully-adjustable
thresholds to determine event detection and release, and
includes the ability to record up to 1 second before the
trigger event.  It can log event times to a separate text file,
but as it stands it doesn't send out any trigger pulses to
external devices.

It writes DQA files, which are standard WAV files with
additional info.  They can be read by anything that follows
the rules for RIFF files (which WAV is a type of), though
you might have to rename the extension to get some
software to consider it.

RIFF files like WAV and DQA are limited to 4 GB due to
32-bit address limits, which for a single 16-bit channel is
about 12 hours continuous recording.  If your events are
infrequent, the total time can of course be much longer
without exceeding 4 GB.

If you really need to record all the background sound
between events, you can't do it with a single conventional
WAV file.  There are a couple of RIFF replacements that
use 64-bit addresses, but you'd be on your own with
those.  The other approach would be to break the
recording into multiple smaller files.

Can you give details of what the trigger is supposed to be
or do?  Does this need to be an external pulse, or some
signal to another piece of software? For the former, I'd
recommend something like Arduino (which Daqarta
supports) that can create 5V on/off states.
<https://www.daqarta.com/dw_rraa.htm>

For pure software, Daqarta can send signals via standard
Windows methods.
<https://www.daqarta.com/dw_ooxx.htm>

The worst-case lag is going to be around 21 msec as it
stands, since the event detector looks at the RMS level
(or instantaneous peak) in 1024-sample chunks.  That
could be improved if needed by using an untriggered
display, which defaults to about 10 ms for waveform or
spectrum, or a spectrogram, which can be set much
faster since it only needs to show one column per update.

Feel free to contact me with any questions.  Of course
you are welcome to try Daqarta; the free trial is 30
sessions or 30 days, after which you can't record from the
sound card.  But you get to use almost everything else
forever, including file analysis, 8-channel signal
generation, psychoacoustics demonstrations, MIDI music
generation, and more.

Tip: Right-click any control to open the extensive Help
system to that topic.

Best regards,

Bob Masta

===================
On 24 Oct 2020 at 10:46, Timothée Proix wrote:

>
> Dear all,
>
> we are looking for some advice to do long audio recordings (potentially a few days). We are
> looking for a system which is able to record voices autonomously and send triggers related to the
> audio (for instance marking the start of the recording), with minimal delays between the input
> signal and the output trigger. The system should also be able to retain large amounts of data,
> because of the length of the recordings. An alternative way would be to implement something to
> do that ourselves.
>
> If you have knowledge of a vendor/technology that could do that, or tips about how to set up such
> a system, we would really appreciate it. We will share on this list the responses we get.
>
> All the best,
>
> Timothée Proix
>