[solved] How to get the actual duration of a sound when "Duration" is set to 0 ms?

Comments

• Josh

  June 2015 edited 11:56PM

  Hi Sabine,

  If you click on your synth item in the overview, you'll see the parameter "Length" above the "Duration" parameter. That value represents how long the tone was played. If you want to control the length per condition, you just determine the value beforehand, in an inline_script or loop item, (e.g. your_tonelength = 150), and then refer to the variable in your synth item, meaning you'd insert [your_tonelength] in the "Length" box.

  Good luck!

  Josh
• SabineS

  June 2015 edited 11:56PM

  Hi Josh,

  Thank you for your reply. I am actually already controlling the "Length" parameter (i.e., I have a [tone_dur] variable specified in a loop and the "Length" parameter of the synth item refers to [tone_dur]. What I want to know is how long the tone was "really" presented. If I would not have the response logging while the tone is still playing, I could simply look at the time when the synth item has finished and subtract from that the start time of the synth item. When I did that, I received deviations from the duration specified in [tone_dur] of about 8 ms (on average). I would like to know whether this deviation magnitude applies also to the current situation, where I do not simply let the synth item finish, but where I move on to an inline script in which I am waiting for key responses.

  Best,
  Sabine
• Josh

  June 2015 edited 11:56PM

  Hi Sabine,

  Did the 'real' time also deviate below your indicated tone_dur? I suspect that there might be an 8ms lag in the initiation of the tone, but not in the tone itself, meaning that the real duration of the tone is always tone_dur. However, if you got deviations below tone_dur this can't be the case; in which case I'd still have to figure out an explanation...

  Cheers
• Edwin

  June 2015 edited June 2015

  Hi Sabine,

  Very good question. There is no programmatic way of knowing the actual sound duration. You will have to use an external device to monitor the sound.

  I would advise to test the system that you use for sound output: have OpenSesame produce sounds of [tone_dur], and record those sounds. You can check the length of each tone in the recording using audio editing software (e.g. Audacity). This can give you an indication of the variability of you sound card.

  The higher quality your computer's sound card is, the less latency there will be in starting and stopping sound playback.

  Alternatively, you could opt for using sound files with a sampler (of which you know the exact length) rather than creating sounds on the spot (with the synth). Sometimes this decreases latency issues (at least within the sound).

  Good luck!

  Edwin
• SabineS

  June 2015 edited 11:56PM

  Hi,

  Thanks for your answers. The deviation between the desired duration and the real duration when using a synth item were both positive and negative (ranging from -9 to 26 ms).

  Regards,
  Sabine
• Edwin

  June 2015 edited 11:56PM

  That's quite a bit of variability; very annoying. Unfortunately, this is an issue of the sound card...

  Have you tried my suggestion using sound files instead of the synth? Does that work better?
• sebastiaan

  June 2015 edited 11:56PM

  Hi guys,

  Just my 2 cents, to confirm what @Josh said:

  I suspect that there might be an 8ms lag in the initiation of the tone, but not in the tone itself, meaning that the real duration of the tone is always tone_dur.

  Indeed, the synth simply generates a tone of the specified duration, and then uses it as a sound file for the sampler. So there is no difference (in this respect) between generating a tone yourself and passing it to the sampler, and using the synth item.

  But because the computer doesn't know perfectly when playback starts and stops, it seems as though the actual sound duration differs--even when it doesn't. If you'd record the sound and analyze it, like @Edwin suggests, it should always be the same length.

  An exception is, of course, when playback stutters. In that case, all bets are off.

  Cheers!
  Sebastiaan

  
  
  Check out SigmundAI.eu for our OpenSesame AI assistant!
  
• Edwin

  June 2015 edited 11:56PM

  Sorry, I must've misunderstood: I thought @SabineS did measure the sound lengths?