Find in the menu.

Use Find to search the sound for silences, peaks, or pop/clicks, or a text phrase in dialogue. Use the Find drop-down list to select what to find. Use the Where drop-down list to specify where to start searching. Select Past selection to find the next occurance. If the finish marker is currently at the end of the file, then the entire file is searched. Enter the search criteria in the boxes provided.

Search Criteria
Find Type Setting Description
Silence Threshold Sets the minimun level of audio. Anything below that is considered silence.
Duration Sets the minimum length of time the audio must remain below the thresold to be considered silence. Using a value of 1.0, for example, ignores silences less than one second.
Peak/clipping Threshold Sets the level of audio that must be exceeded to be considered a peak. Anything below that level is ignored. Use a thresold of 0dB (or slightly below) to find areas of clipping before saving a file.
Duration Sets the minimum length of time the audio must remain above the threshold to be considered a peak or clipping. Usually this value will be 0 or very small, spanning just a few samples. For a 44100Hz file, a value of 0.00007 would be about three samples (3/44100).
Pop/click Tolerance Sets the percentage of change in the waveform that is to be considered a pop or click. Usually pops or clicks have a very sharp change in the waveform of about 700% or more. Start with a higher value and decrease it as required to avoid finding false positives. Some solo instrumentals (drums, trumpet, etc.) naturally have sharp changes where reliable pop/click detection is not always possible.
Text Phrase Sets the search text. This can be a single word, but several words usually give better matches. Everything should be spelled out, including numbers and symbols. Enter "twenty four" instead of "24" or "plus" instead of "+".
Recognizer Sets the speech recognizer. On most systems only the "Microsoft Speech Recognizer" is available.
Confidence Sets the confidence level for recognition. If you find that too many false matches are found, increase the confidence value. A value of 100% requires a very high level of confidence for each match. However some matches may be missed. Using a confidence below 50% usually results in too many false positives.

Speech recognition rarely works well with music or singing. Ideally the audio should contain dialogue only with low background noise. Use to minimize the background noise or use to reduce the music and extract the vocals. Due to the limitations of current speech recognition technology, false positives or missed matches may still occur.

Use the Zoom in on selection if found checkbox to automatically zoom in (up to 1:1) on the selection when found. If this box is not checked, then the current zoom level is used and the view is scrolled to the selection instead.