The transcription toolbar offers some simple tools that are useful for transcribing recording into text and text labels, automatically finding the beginnings and endings of isolated speech.
Most of the functions work by using a sound detector, which detects the RMS of a region of sound. (Other cababilities of the detector include detecting changes in zero-crossings or direction-changes, but these do not currently work. Future versions may enable these).
To use the sound detector, it first must be calibrated to noise, so that it can determine what tolerance you want. This is done by selecting a region of relative silence (an area that doesn't contain the noise you wish to detect) and pressing the calibrate button (fifth from the left).
Once calibrated, you can adjust the sensitivity of the sound detector with the slider on the right side of the toolbar. Farther to the left will consider small differences to be real, and farther to the right will only consider larger differences to be real.
The four buttons on the left adjust the selection region by moving the selection boundaries inward. The two leftmost buttons move the outer boundary inward from a silent region to a region of sound. THe next two buttons move the boundary inword from a sound region to the next silent region. Together, these buttons can be used to adjust the selection 'automatically' to the sound detector's best guess for where certain sounds start and end.
Once a region is selected, a label on the labeltrack can be created by using the make label button (rightmost button). This is also accessible with the the CTRL-B shortcut.
Finally, if the sound detector is calibrated and the sensitivity is set to an acceptable level, the remaining (second-from-the-right) button will automatically detect the beginning and ending of continuous sound regions and create labels from them. This is fairly slow, and may not be reasonable for detecting the start and end of tracks on a record, but it may be OK.