In this study, algorithms of pitch measurement were compared in order to define whether existing algorithms can provide reliable results for digital recordings of speech produced by oesophagus. Due to properties of that type of speech, results were not always predictable. Several well-known pitch detection algorithms were selected for the research. Experiments show that most of the pitch detection methods were not robust enough to provide reliable results for all speech samples. Auto-correlation methods and spatial pitch network model failed at one of the samples, each on the different one. Performance of spectral subharmonic summation pitch detection for the given samples was more reliable, though with a high dispersion of the results. The YIN method had failed all the attempts for one of the samples, getting confused while distinguishing non-harmonic part from harmonic. We can infer that time domain approach for pitch measurement is more suitable for working with the esophageal voice, as it has lower dependence on the shape of the waveform. However, time domain algorithms need to be revised in order to produce reliable results for hoarse and noisy esophageal speech signals, e.g. by applying additional filters or means of detection of harmonic part of the speech signal.