Step 3 - Temporarily DownSample Your Wav File

Forced Alignment using HVite only seems to work with audio recorded at a 16kHz sampling rate, at 16bits per sample.  So we will create a temporary version of your audio at 16kHz/16bits, run HVite, and then use the time alignments generated by the Forced Alignment process to segment your original audio.

You need to use the SoX sound editing utility to downsample your audio.  The SoX command syntax is as follows:

    $ sox original.wav -c 1 -r 16000 -w downsampled.wav 

• -c 1 - converts stereo audio to mono
• -r  16000 - is the target rate of 16kHz
• -w is the target bits per sample (word length = 16 bits)

In this example, we will use the audio from jimmowatt's "History of England" submission:

 historyofengland01ch04_01_macaulay.wav 04-Mar-2007 14:32   165M  

Which we will downsample to 16kHz-16bits using the following sox command:

    $ sox historyofengland01ch04_01_macaulay.wav -c 1 -r 16000 -w downsampled.wav

which creates the downsampled.wav file.   It's a good idea to open the resulting file with Audacity to make sure the file converted OK.