Sound Activated Recording, not very useful

[autark]

I have tried to use Sound Activated Recording a few times but have always ended up disappointed. In my experience, the deactivation is over-eager, optimistically stopping the recording whenever there is the least bit of silence. This makes the function unsuitable for recording dialogue or music containing silent passages.

I don't know, maybe fiddling with the cutoff level could work around the issue somehow, but it would be a lot more useful if the deactivation would only kick in after a user-settable grace period, and the default should at least be some five seconds or so. I don't think anyone would mind having to trim a few seconds off the end of a recording, compared to the risk of having a chopped-up recording.
--
Cheers,
Per Å.

[steve]

There are already a couple of "feature requests" for improving Sound Activated Recording, but what exactly is it that you are trying to do? There may be some subtle differences between the improvement that you would like to see, and those already suggested, or there may be some technique that may already allow you to accomplish your task. Details are really useful.

[autark]

OK, my use case is this:

I want to record from streaming audio. I want the recording to start when I start the audio, and I want the recording to stop automatically when the audio is finished. The audio contains a number of songs, each separated by a couple of seconds' silence. There is also some dialog, with some moments of silence. I want to record the audio as-is, conserving all silence expect at the beginning. I can accept some silence at the end of the recording.

My experience with this use case and Sound Activated Recording is that the recording is activated correctly, but any time the sound level runs below the sound activation level the recording is paused instantly, effectively condensing all silence into almost nothing.

What I'm asking for is the automatic pausing of the recording to not occur until the perceived silence has been going on for a period of at least X seconds. The period should be configurable with a sensible default (5 or 10 seconds seems reasonable). Just to be perfectly clear: I'm happy with the recording activation, it's the deactivation/pausing logic that is much too eager.

I hope I'm making myself clear.
--
Per Å.

[bgravato]

I totally agree with Per. I think this has been already discussed and posted in feature request forum, wasn't it?

That and automatic labels with timestamp every time sound activated recording is started... the two missing 'killer' features in sound activated recording

[steve]

The current entry on the Feature Request page of the Audacity wiki has the following:

* Sound Activated Recording (currently in Audacity Beta only)
o Pre- and post-roll duration: define a duration to be recorded before the threshold-level sound occurs, and similarly a duration to be recorded after the input falls back below the threshold. Requires buffering of the recording for pre-roll, but is expected behaviour for this type of tool. (27 votes)
+ A post-roll delay (say 4 seconds) would allow realistic recording of dialogue
o Segment each recording event (13 votes)
+ with labels (12 votes) - labels could contain either the length of preceding pause, or date and time when recording started/resumed
+ with clip lines (1 votes)
o Report system clock times for each recording event: (6 votes) could be a display on timeline; added to names of labelled regions; available in Debug log; or an exported text file via a button in Recording preferences (data only retained per session)
o Adjustable start-time (2 votes)

This is already possible if you turn on Sound Activated Recording and then start Timer Record. You just have to set an appropriate duration (not end time, obviously ) so that recording doesn't stop before you want it to

There is also a page started on the wiki specifically about "Voice Activated" recording: http://wiki.audacityteam.org/wiki/VoiceActivated
This seems to be exactly what you are asking for, so I've added your vote.

I think that "pre-roll/post-roll" would also achieve the result that you are asking for, so would you like me to add your vote there too?
(You can register on the wiki and add your own vote if you prefer).

For your specific user case, I think that you may be able to achieve the desired result by setting the activation threshold level really low. When streaming audio playback ceases, the sound level should drop to very near silence as long as all other audio inputs are muted, thus it should be possible to set the silence threshold higher than the quietest parts of the streaming audio, but higher than the noise floor when the playback stops.

[autark]

@steve: Thanks for adding my vote. Pre-roll/post-roll sounds fine.

[autark]

For your specific user case, I think that you may be able to achieve the desired result by setting the activation threshold level really low. When streaming audio playback ceases, the sound level should drop to very near silence as long as all other audio inputs are muted, thus it should be possible to set the silence threshold higher than the quietest parts of the streaming audio, but higher than the noise floor when the playback stops.

Yes, you're right. Setting the level to -59 dB seems to work for streaming audio.