A tool for recording lists of individual words as separate audio files. WordRecorder automatically listens for speech and trims the resulting audio file accordingly.

Version 1.0
By Doug Letterman


WordRecorder is a specialized tool to aid in the recording of a large set of individual spoken words as separate audio files. It was originally created for a video e-card site where each card would be personalized with a professional announcer reading the name of the recipient. To make this possible, the site developers needed a method to record hundreds of names as separate audio files for later inclusion in e-cards on demand.

WordRecorder works by accepting an input file that contains a list of words or phrases, one per line. The words from the file appear in a list to the right of the main interface. The user can record each word individually or hit "Record All" to record every word in the list in series.

Operating System

Currently WordRecorder only works on Mac OS X (it uses Apple's CoreAudio Framework). The interface and everything else is written with Qt, though, so eventually it shouldn't be too difficult to port WordRecorder to other operating systems.


The latest release and source code of WordRecorder may be found here:

The SourceForge project is here: http://sourceforge.net/projects/wordrecorder.


The main window

The Word List

The first step is to load the word list file using the “Open Word File” button or “File > Open...” menu item. WordRecorder treats each line of the input file as a separate "Word", regardless of how many words are actually on a given line. As a result the input file could contain whole sentences, phrases, or groups of words separated by spaces. There is one important limitation to this scheme, though. WordRecorder will name the resulting recorded audio file using the string of characters before the first space in the line. For example, the line:

“Grand Central Terminal”

Would result in an audio file called Grand.aiff. It is important, then, that no two lines in the word file start with the word ‘Grand’.

Audio Files

Before recording set the destination directory for the resulting collection of audio files using the browse button at the lower right hand corner of the recorder panel.

Once a destination has been chosen, pressing the “Record One” or “Record All” buttons will prompt WordRecorder to start waiting for non-silence (see “Record Threshold,” below).

After one or more audio files has been recorded check marks will appear next to their names in the word list. Selecting a word from the list and hitting the play button will play back the audio file (Note: The VU Meter does not display the audio levels of the file during playback).

audio files

WordRecorder outputs audio files with the name of the current word from the list.

The options dialog

Setting Options

WordRecorder's options dialog, “File > Options”, allows the user to fine-tune the automatic recording features, audio output format, and audio file trimming characteristics of the program.

Record Threshold
This is the audio level at which the WordRecorder will start recording, measured in dB. Any audio peak above this level will trigger the record —it does not have to be a sustained average.
WordRecorder will stop the recording after the average audio level (not the peak) falls below this value for the duration of time indicated by “End Padding” (see below).
Begin Padding
Due to the nature of the human voice, if WordRecorder were to start the recording exactly when the audio levels first reach the indicated peak level it would likely chop the word off before the first syllable. This setting indicates the amount of buffered audio from before the peak level has been reached that will be copied into the head of the resulting audio file.
End Padding
This setting serves two purposes. It indicates both the duration of below-level or silent audio that will be placed on the end of each audio file. It also represents the amount of time that the average audio level must stay below the record threshold before WordRecorder stops the recording. These two purposes should probably get their own settings values in a future version of WordRecorder.
Audio Format
The format of the output audio file. Currently WordRecorder supports AIFF or m4a audio.

Known Limitations

  1. The End Padding setting is also used by the WordRecorder to determine how long to wait after not hearing any non-silence before stopping recording. It would be desirable to have this broken out as a separate setting, allowing users to trim audio more closely at the tail while permitting longer intervals of silence during the recording itself.
  2. Currently the VU Meter only meters the incoming audio signal. It does not meter the levels of audio during file playback.
  3. In a future version the VU Meter should display the current record threshold.
  4. WordRecorder should support the selection of an audio input source other than the system default
  5. The VUMeter currently only monitors the left input level


WordRecorder is distributed under the terms of the GNU General Public License.

$Id: index.html 9 2008-04-30 15:49:34Z dletterman $