"In general, my process consists in obtaining the tracing of the to-and-fro movements of a vibrating membrane, and the utilization of this tracing for reproducing the same to-and-fro movement, with their relative inherent durations and intensities in the same membrane, or in another adapted for furnishing the sounds and noises which result from this series of movements.
"We are, therefore, concerned with the transformation of an extremely delicate tracing, such as that obtained with a delicate stylus rubbing upon a surface blackened by a flame, to transform, I say, these tracings in relief or intaglio, in resisting material capable of guiding a moving body, which transmits these movements to the sonorous membrane.
"A light stylus is connected with the centre of a vibrating membrane; it terminates in a point (metallic wire, the barb of a feather, etc.), which bears upon a surface blackened by a flame. This surface is a part of a disc to which is given a double movement of rotation and rectilinear progression.
" If the membrane is at rest, the point will trace a simple spiral; if the membrane vibrates, the traced spiral will be undulating, and these undulations represent exactly all the to-and-fro movements of the membrane, with their times and intensities."
Up to this point the apparatus as described would represent a modified Scott phonautograph in which the cylinder is replaced by a flat disc. M. Cros then continues:
"By means of the photographic process which, in fact, is well known, this traced, transparent, undulatory spiral is converted into a line of similar dimensions, in intaglio or in relief, in resisting material like tempered steel, for instance,
"This done, this resisting surface is, by means of a motor apparatus, made to turn and to progress rectilinearly with a velocity like that which was used in the registration.
"If the reproduced tracing is in intaglio, a metallic point (and if it is in relief, a notched finger), held by a spring, bears upon the tracing at one end, and is connected at the other end with the centre of the membrane adapted for sound reproduction. Under these conditions, this membrane is not any more acted upon by the vibrating air, but by the tracing, controlling the pointed stylus by pulsations exactly like those to which the membrane was subjected in recording, both as to duration and intensity."
This paper was only read in open session at the Academy on December 3, 1877. Nevertheless to Charles Cros belongs the honor of having first suggested the idea of, and feasible plan for, mechanically reproducing speech once uttered. (text written by Edmunds in 1888 with quotations from Cros paper)
Source: "The Graphophone" by Henry Edmunds, paper read 7 September, 1888, at Section G of the British Association for the Advancement of Science, Bath Meeting, copy in the Tainter Papers, Smithsonian National Museum of American History, Washington, D. C.