What are the difference between speech and music signal? Can anyone help?

I am doing project in audio processing. In that project I want make decision, that is which type of signal is given as input whether it is speech signal or music signal. I searched a lot in internet. But i could not find the required details.

If anyone knows about this please share the difference between the speech and music signal.

  • 2
    $\begingroup$ is there always a difference? when someone is singing, what is the difference between speech and music? $\endgroup$ Mar 5 '15 at 15:09
  • $\begingroup$ I am not here to answer. I have a question? When two music signals played by instruments are binded to form a one signal there will be change in frequency . Can we retrieve two signals back with the same frequency given as inputs ?. Your reply will be very useful for my project?. $\endgroup$ Jul 18 '17 at 16:02
  • $\begingroup$ @AnanyaLakshmi : You should ask a question. Answering a question with a question (as you did) is bad form for this site, so I have converted your "answer" to a comment. $\endgroup$
    – Peter K.
    Jul 18 '17 at 17:00

Not really an answer, but usually when people speak there are pauses between words, sentences, thoughts, etc. Music (with multiple instruments) is fairly continuous. Even if one instruments stops another is probably still playing. So there would be less lulls in the audio stream. Just a guess though.

Another thing you could do is look at frequencies, voices are typically in the 300Hz-4kHz range (this is what telephones originally used as bandwidth) whereas musical instruments are usually much broader. A frequency outside the human range could indicate instruments enter image description here I randomly found that image here


There's probably some overlap (some people seem to talk in a drone or sing-song), but the time and pitch distribution of music is likely to contain more pitch or voiced frequencies that are spaced apart by chromatic intervals (at least for Westerners), and more abrupt changes of pitch spaced apart by close to close to some multiples of some fixed tempo interval, compared to the speech of the typical radio news announcer. e.g. a compressed spectrograph will show a bit more vertical and horizontal checkerboard patterning for most music than for most speech.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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