ORIGINALLY PUBLISHED LIMA NEWSLETTER OCTOBER 1988 ^^^^^^^HOW ACCURATE IS THE 99/4A'S SOUND CHIP? ^^^^^^^^^^^^^^^^^^by Charles Good Some of the musical programs in our user group's library sound just a bit off, with an occational sour note. This is particularly true of songs played as single notes rather than chords. You don't have to be a music expert to detect these sour notes. Either the programmer programs in the wrong frequency, or perhaps the sound chip is less than totally accurate. Reading about the CALL SOUND statement in the Users Reference Guide makes one believe that the sound chip is very accurate. The frequency portion of a CALL SOUND statement can have a frequency between 110 and 44733 hertz (cycles per second) suggesting that there are 44623 separate sounds possible! Of course, many of these frequencies are beyond the ability of most monitor speakers or exceed the limits of human hearing, but these numbers suggest the possibility of great accuracy. The User's Reference Guide on page III7 gives the frequencies for four octaves of of commonly used musical notes. In order to check the accuracy of the musical frequencies listed on page III7 I wrote the following program: 100 PRINT "INPUT FREQUENCY" 110 INPUT F 120 CALL SOUND(4000,F,0) 130 GOTO 100 I entered various frequencies, including those on page III7 and then used an ARION HU8000 Micron Chromatic Tuner to measure the resulting sounds as they came out of the speaker on my TI 10 inch color monitor. This chromatic tuner is used by musicians to tune stringed instruments, including pianos, and is quite accurate. I played around with different CALL SOUND frequency values and found those that, according to my Chromatic Tuner, were closest to being perfectly "in tune" for all of the notes listed on page III7. The results are shown in the table below. These data were identical on each of the two 99/4A consoles I checked. Not all frequencies listed on page III7 actually produce the desired note. These are noted in the COMMENTS section of the table. The inaccurately published frequencies are in fact the theoretically correct values, but they are one hertz too large to give the correct sound from the TI sound chip. Music programmers take note! This may be the source of some of the "off" sounds in some of the songs in our group's software library. It is possible that these frequency errors are due to the speaker used in the test rather than the TI sound chip. Maybe a high quality speaker would make these errors disappear, but I doubt it. For some notes, it is not possible to get a sound that is perfectly in tune. This is particularly true of the higher notes. Values shown in the table show which CALL SOUND frequency gives the closest approximation of a note, and the percent error from perfect "in tune". Again, a beter quality speaker might make these apparent errors disappear, but I doubt it. It is possible, however, that the magnitude of the high frequency errors (10-15% off true "in tune") would be reduced with a better speaker. Beginning with E above middle C (330 hertz) more than one adjacent frequency number produces EXACTLY the same sound. Thus the apparent tremendous accuracy of the TI sound chip with its 44623 different sounds is a mith. The number of possible sounds is much less. This particular characteristic of the TI sound chip would not be affected one way or another by a higher quality speaker. For A above high C (theoretically 880 hertz) any frequency number between 878 and 884 in a CALL SOUND statement produces the EXACT SAME in tune sound. If you go one higher to 885 you get a distinctly higher sound. You don't need a sound meter to detect this change of sound between 884 and 885. The difference can easily be heard with normal hearing. The above small BASIC program lets you input a new frequency before the old one quits playing so you can listen to frequencies back to back with only an instantaneous pause in between. Try it. Type in the program and RUN it. Input frequencies 878, 879, etc up to 884 and they all sound EXACTLY the same. Then try 885 and notice the difference. Hopefully the data in this article will help music programmers to appreciate some of the limitations built into the TI sound chip and correct some of the sour notes published on page III7 of the Reference Guide. NOTE |CALL SOUND |Actual measured| COMMENTS |frequency, |CALL SOUND | |as reported. |frequencies | |on page III7 |that generate | |of Users Ref.|this note in | |Guide. |tune. | ----------------------------------------------------------------- A........110.............NA.....Volume too soft to meter. A#.......117.............NA..... " " " " " B........123.............NA..... " " " " " C(low C).131............131 C#.......139............138.....Off by 1HZ, also 5% flat D........147............147 D#.......156............156.....5% sharp E........165............165 F........175............175.....5% sharp F#.......185............185 G........196............196 G#.......208............207.....Ref Guide high by 1HZ A........220............220.....5% sharp A#.......233............233 B........247............247 C(mid C).262............261.....Ref Guide high by 1HZ C#.......277............277 D........294............293.....Ref Guide high by 1HZ D#.......311............311 E........330..........329-330 F........349..........349-350 F#.......370............369.....Ref Guide high by 1HZ G........392..........392-393 G#.......415..........415-416 A........440..........440-441 A#.......466............466 B........494..........491-491...Ref Guide high by >=1HZ C(high C)523..........522-523 C#.......554..........553-555 D........587..........584-587 D#.......587..........584-587 E........659..........656-659 F........698..........697-701 F#.......740..........738-743 G........784..........780-784 G#.......831..........826-831 A........880..........878-884 A#.......932..........929-935 B........988..........986-994 C.......1047.........1041-1050 C#......1109.........1102-1112 D.......1175.........1172-1183..5% sharp D#......1245.........1236-1249 E.......1319.........1309-1323 F.......1397.........1389-1406 F#......1480.........1462-1481..10% flat G.......1568.........1565-1586..5% sharp G#......1661.........1657-1681..10% sharp A.......1760.........1734-1761..15% flat