One of Helmholtz's most celebrated devices is the apparatus for reproducing vowels synthetically by means of tuning forks. (90) It consists of ten electrically driven tuning forks in connection with adjustable resonators, Fig. 179. Helmholtz describes the experiments by which he says he produced a very find no, a good gloom, and a passable raw, while the other vowels could be only imperfectly imitated. Zahm, who repeated the experiments with the Helmholtz apparatus, remarks that the resemblance of the artificial sounds the the natural ones is, at best, more or less fanciful. Rayleigh says: "These experiments are difficualt and do not appear to have been repeated."

Helmholtz explains the difficulties as in part due to the fact that the higher forks give only weak tones and that the series is not large enough; his highest fork gave 1792 vibrations per second. Other difficulties are that Helmholtz and others have not known the exact composition of any of the vowels; and, had the composition been known, there was no adequate method of adjusting the several components to the proper intensity. In order to imitate an actual vowel, it is desirable that the pitch of the fundamental shall correspond exactly to that of the voice being reproduced. Helmholtz had only one series of forks giving eight partials based on 112 vibrations per second, which was later extended to give also eight partials based on 224 vibrations.

Helmholtz also tried organ pipes, and says: "We can effect our purpose tolerably well with organ pipes, but we must have at least two sets of these, loud open and soft stopped pipes, because the strength of tone cannot be increased by additional pressure of wind without at the same time changing the pitch."

Our study of organ pipes showed that these difficulties are not insurmountable and that pipes can be made which are more advantageous than tuning forks for experiments in synthesis.

A vowel having been photographed and analyzed, the synthesis can be performed in a strictly quantitative manner by means of the phonodeik. A set of pipes is prepared, one pipe for each partial of the given vowel; for the vowels as spoken by D C M the least number of pipes is six for the vowel gloom, while the vowel father requires ten pipes, and the vowel mat sixteen; the group of pipes for the latter is shown in Fig. 180.

Even sixteen pipes reproduce only the more important partials of the vowel mat, since the full analysis shows twenty or more components in some instances. For voices of higher pitches the number of pipes required is less. Each set of pipes is mounted as compactly as possible on a separate, small wind chest. The smaller pipes alternate with the larger ones and their mouths are on a different level to prevent interference.