Rockbox mail archive
Subject: RE: Silence detection
From: Andreas Stemmer (groovingandi_at_gmx.de)
after some testing I can finally give some comments on the silence detection
> Do you think the trigger modes (off, no rearm, rearm) have good names?
> Are they understandable easily?
Perhaps not easily, but understandable. You have already changed them to
"off, one, repeat",
which are better names for the trigger modes in my opinion, too.
Just one cosmetical correction perhaps: at the moment it's "Off, one,
repeat", it would be nicer to have "Off, One, Repeat" in order to be
> I still feel a bit uncomfortable about the optical feedback of the
> trigger state too. I tried to visualize the trigger states and
> thresholds in the peakmeter. Is it understandable what the peak meter
> displays? (You might want to try with minimum trigger durations > 0
> to see all states.)
Again, it's understandable if you think about it, but it's not
self-explaining enough. Would it be possible to visualize the thresholds
with vertical bars in the peakmeter? The trigger states don't have to be
displayed in the peakmeter, a simple line of text would be enough for that.
At the moment there's "trigger idle" and "trigger active" which was a little
bit confusing for me at the beginning. "trigger active" could also mean that
the trigger is waiting for the threshold which is "trigger idle" at the
My suggestion would be "trigger armed" and "triggering" (perhaps with a
countdown) in an extra line, not in the size-line. I don't really need three
lines for the gain settings on the recording screen, one could use two of
them for displaying the trigger status.
Btw: the size of the recorded file is never reset, whereas the time is reset
to 0. This is a little bit confusing.
> Are the values for the minimum trigger time
I can't imagine any scenario where I could use the 10min setting or a
setting which is bigger than the prerecord buffer, but it doesn't hurt..
> Do you think it is a good idea to put the prerecord time
> onto the trigger setup screen, too?
Yes of course, triggering only works reasonably with a prerecord time
adjusted to the minimum trigger time.
> If so - should the menu item for
> the prerecord time vanish from the recording menu?
No, because the prerecord feature is useful for normal recordings too, and
nobody would search it in the trigger menu.
> Assume you have the intension to start a recording with a trigger but
> stop it manually. Is it sufficiant that the user can set a stop
> threshold to 0% (-89db) (maybe displayed as "off") or is it necessary
> to put that into an additional trigger mode?
> The same applies when you want to start a recording manually but want
> it to stop automatically on silence. (only that start threshold 100%
> (0db) is equivalent to off)
I think an "Off"-Setting is enough ("off" is better than "-inf" because its
easier to understand for non-techie users)
One thing I noticed: if I have a prerecord time of 3s and the minimum
trigger set to 1s, I expect to get a recording with 2s of silence below the
threshold. What I get is a little bit more than 1s of silence (1.1s in this
case). At the end (with setting 1s), I get a little bit less than 1s (0.7s)
of silence. I don't know if it is possible to adjust the timing so that you
get the expected time of silence in the recording?
One last remark to the automated splitting (activate split as it was called
by Martin Borus): I don't see, why there should be three different threshold
and I didn't understand it completely yet what it should be good for. If you
want to record all the songs from a tape and use the trigger feature as it
is at the moment, you'll probably lose some of the silence between the
songs... but where should the split point be if you want to keep the
silence? Why is it necessary to keep the silence?
The approach I would prefer is: introduce some kind of "paranoia"-mode which
records all the time and starts a new file (frame-accurate) on each split
condition. This could be useful for important recordings. If I want to
record a live set of our band, I don't have time to monitor the levels all
the time. With this paranoia-mode, everything would be recorded. If I have
set my thresholds correctly, I get one file for each song and one file for
the silence between the songs which can be deleted afterwards. If the
thresholds were not optimal, it would at least be possible restore the parts
which were accidently marked as "silence".
While writing this, I had another idea of a completely different approach:
Why not simply record all the time and create a cue-file in parallel
according to the trigger thresholds? Splitting or cutting can then be done
afterwards without danger...
Of course, this way would not be very good for applications like
surveillance recording (monitor a room for example for days and only record
if something happens...)
I hope this helps a bit,
Page was last modified "Jan 10 2012" The Rockbox Crew