We have reported previously (San Francisco ASA meeting in November 1975) on the intelligibility and discriminability of a low‐frequency formant speech does with both normal hearing and hearing impaired subjects. Since then we have endeavored to discover just how much degradation the speech suffers in the process of coding and at what stage of the coding it occurs. Using Form IV of the Diagnostic Rhyme Test as test material, we have presented a modified speech code (dealing with partials of the speech rather than formants as such) at various stages of coding to normal hearing subjects. (This was done to assess the adequacy of the code and to initially avoid training hearing impaired subjects.) The test is used in the usual “intelligibility” mode in which the subject hears a single stimulus and is asked to identify it as one member of a word pair for the following conditions: (1) speech directly as recorded, (2) speech low‐pass filtered (4.5 kHz) followed by analog‐to‐digital and digital‐to‐analog conversions, (3) speech synthesized from largest amplitude partials. In addition, the test is run in a “discrimination” mode in which the subject hears a pair of rhyming stimuli and is asked to judge whether they are the same or different for the above conditions and for conditions after low‐frequency transposing by a factor of four to one. Work is in progress with hearing impaired subjects.