General Discussion: Summary, Error Analysis, Applications

Abstract

In the preceding chapters, the prevailing methods of pitch determination have been presented and discussed. The different principles of operation as well as their individual advantages and limitations have been demonstrated as far as possible. In this final chapter, our attention is focused on some important issues dealing with pitch determination from the speech signal in general, such as:

• Has the problem, in general as well as from the point of view of the respective application, been solved? If not, which basic questions remain?

• What are the special advantages of the individual principles? Once a special advantage has been found, is it a matter of implementation (and thus transferable), or is it intrinsic to the particular method?

• What are the weaknesses of the individual principles? Are they a matter of implementation, or are they intrinsic?

• If an individual PDA or a class of PDAs fail temporarily, do they fail with respect to signal, range of F₀, distortions, noise, spurious signals, or recording conditions? If such a failure is intrinsic, is there another PDA which might be insensitive to this particular situation?

• Which PDA is best suited for which application?

• What requirements with respect to output and display facilities arise from the individual applications, and how are they met?

• What arguments concerning pitch determination principles will gain importance, and what arguments will become obsolete under the impact of modern large-scale integration technology?