Robust pitch period detection using dynamic programming with an ANN cost function

S. Harbeck, Andreas Kießling, Ralf Kompe, Heinrich Niemann, Elmar Nöth

In this paper, a new pitch synchronous FO-algorithm is described. The task of detecting pitch periods in the speech signal is solved with a search for an optimal path through a space of pitch period hypotheses. The search is efficiently implemented by dynamic programming (DP). The DP cost function is computed with an automatically trained artificial neural network (ANN) which combines the outputs of heuristic functions measuring the similarity of adjacent period hypotheses. With this algorithm a coarse error rate of 4,75% on a German speech database is achieved. It outperforms the DPFO algorithm, which itselfs outperforms two "conventional" algorithms.

doi: 10.21437/Eurospeech.1995-255

Cite as: Harbeck, S., Kießling, A., Kompe, R., Niemann, H., Nöth, E. (1995) Robust pitch period detection using dynamic programming with an ANN cost function. Proc. 4th European Conference on Speech Communication and Technology (Eurospeech 1995), 1337-1340, doi: 10.21437/Eurospeech.1995-255

@inproceedings{harbeck95_eurospeech,
  author={S. Harbeck and Andreas Kießling and Ralf Kompe and Heinrich Niemann and Elmar Nöth},
  title={{Robust pitch period detection using dynamic programming with an ANN cost function}},
  year=1995,
  booktitle={Proc. 4th European Conference on Speech Communication and Technology (Eurospeech 1995)},
  pages={1337--1340},
  doi={10.21437/Eurospeech.1995-255}
}