Help Me With My Thesis Project Please!

[sonicfury]

Hello, I am currently conducting my Master's thesis experiment, and it involves the recording and analysis of vocal responses made by subjects.

In the experiment, one of four sessions is carried out for 12 minutes, during which Audacity is recording everything. A stimulus is played through speakers in front of the subject (the microphone picks this up), and then it is the subject's task to respond to that stimulus vocally. This happens about 50 times each session, with varying amounts of time in between each presentation.

I need an efficient method to determine the exact elapsed time from the onset of a stimulus and the onset of the subject's vocal response. In other words, the subject's reaction time.

Is there an easy way to have these intervals automatically extracted for me? Maybe I should be looking into removing the silence between each task, at the very least to narrow my search time through the audio file?

Or maybe there's a better program for this? My lab has funds, so buying a program is not out of the question.

Thank you SO much for any help you can offer, this is extremely important to me!

Dusty

[Irish]

Upgrade to Audacity 1.3.11, available here: http://audacityteam.org/download/

1.3 has a function called "Sound Finder" (Analyze > Sound Finder) which will place labels on the regions where there are sounds beginning after silence. You can choose the level of the threshold between sound and silence.
(1.3 has the concept of "Region Labels", which 1.2 doesn't have)
1.3 also has a much better time display than 1.2, and if you select the portion of audio from the beginning of one label to the beginning of the next, it will tell you how long it is.

PO'L

[kozikowski]

We usually groan when the question uses words like: "How can I automatically..."

Audacity isn't big on automation, scripting, or batch processes, although it's getting better.

If you have the stimulating sound and the response on the same timeline, it's certainly possible to manually derive the silent gap. Change the time window on the bottom of the work panel to "Length" and select the gap.

None of the fancy tools are available in Audacity 1.2. Download and use Audacity 1.3.11. If you're going to get any automation or batch processing at all, that will be the way to get it.

You can put both 1.2 and 1.3 on your machine as long as you only use one at a time. Audacity 1.3 projects will not open in 1.2.

Koz

[sonicfury]

So far, your suggestions sound like they're exactly on track with what I'm looking for.

I'll give it a try and let you know if things don't quite go like they should.

Thank you SO much, this is a huge help!

[kozikowski]

<<<Is there an easy way to have these intervals automatically extracted for me? >>>

How automatic did you want that? Silent Sense or Music Sense is traditionally hard to adjust, particularly since a controlled, quiet room is completely beyond most people. The room will always have background sound, sometimes quite loud.

There is the joke of software that can find the "quiet" regions between cuts on your phonograph records. The record that needs this to work the most will always fail because popular records are played a great deal leading to noise between the songs. There is no silence.

So yes, those tools will place the labels, but you still need to manually manage the data and clear all the mistakes.

Koz

[sonicfury]

Alright, I have the beta downloaded and have spent some time with it. The new time readout at the bottom is pretty nice, great call there.

After some toying around with the Sound and Silence finders, I've managed to make them both put markers in the appropriate places. Where can I go from here? Is there an easy way to delete all the space between the sound markers?

[kozikowski]

Here is a reading from the Audacity 1.3 instruction book.

http://manual.audacityteam.org/index.ph ... bel_Tracks

The only other thing I could find was a development page for Labels.

http://wiki.audacityteam.org/wiki/Label_Track

Koz

[steve]

File menu > Export Labels

This will create a plain text file like this:

Code: Select all

0.000000	2.561779	label 1
2.561779	5.444811	label 2
6.705107	8.772652	label 3

The first number is the start of the label region, the second number is the end of the label region, the word(s) at the end is/are the label text.
There is one label per line.

[kozikowski]

Are the numbers in seconds?

Koz

[steve]

<<Are the numbers in seconds?>>
Yes.