The speech file should be recorded at 16 bits/sample, 16000 samples/second, and named "pre_g722.wav".
The ITU tests use the codec in a special mode, in which the QMFs, which split and recombine the sub-bands, are disabled. This means they do not test 100% of the codec. This is the reason for including the additional listening test.
The files, containing test vectors, which are supplied with the G.722 specification, should be copied to itutests/g722. The ITU tests can then be run by executing g722_tests without any parameters.
To perform a general audio quality test, g722_tests should be run with a parameter specifying the required bit rate for compression. The valid parameters are "-48", "-56", and "-64". The file ../test-data/local/short_wb_voice.wav will be compressed to the specified bit rate, decompressed, and the resulting audio stored in post_g722.wav.