I try to have a signal (speech) represented in time and frequency domain (waveform, spectrum and spectrogram). I use Sonic Visualiser and for spectrum representation it shows an analysis at a given point in time. I wonder what representation should I use to represent a selected area of the waveform e.g. a word, in frequency domain (spectrums). Sonic Visualiser doesn't give averaging options.


  • $\begingroup$ Wavesurfer does with LTAS (Long-Term Average Spectrum) $\endgroup$
    – jojeck
    Apr 8, 2015 at 8:52

1 Answer 1


It would not be useful to average in time because vocalisations are represented directly by their spectrogram. (For example: http://www.u.arizona.edu/~ohalad/Phonetics/notes/Formants%20Spectrograms%20and%20Vowels.PDF or even http://www2.coloradocollege.edu/dept/ev/Research/Faculty/OVALItems/FordOsborneVocabulary/_SouthernVocabularyTable.html)

I would suggest that you export your spectrogram as an image and then isolate and layout the areas you are interested in, in some other software like GIMP. For more information, please see: http://www.sonicvisualiser.org/doc/reference/1.9/en/#imageexport and http://www.gimp.org/

Hope this helps.

  • $\begingroup$ Thank you A_A for your answer, and specially for the links. Jojek thanks a lot, I think I will use it. $\endgroup$
    – Vasiliki
    Apr 8, 2015 at 10:05
  • $\begingroup$ You are welcome, if you feel that the answer helped you, you can upvote it via the counter arrows on the left. $\endgroup$
    – A_A
    Apr 8, 2015 at 13:21

Not the answer you're looking for? Browse other questions tagged or ask your own question.