In Reply to: RE: You can tell by using a computer audio editor posted by rick_m on September 24, 2008 at 19:44:35:
While it's easy to compare two versions of the same recording, it would be much more interesting to be able to tell from looking at a waveform whether the music would sound better one way or the other. Unfortunately, there are conflicting clues in waveforms: initial transients, positive/negative peaks, periodic asymmetric waveforms all provide clues, but they are not consistent, and I have found no simple way to tell which polarity is going to sound best except by listening (and even then it is not always reliable). I didn't try to write any DSP software. Perhaps it could be done.
There has been some work on automatic recognition of speech polarity. This is an easier problem than a typical music recording which has lots of reverberation and multiple instrument types, some of which have no preferred orientation. (For example, bipolar radiators such as kick drums will have different polarity according to their orientation to the microphone.)
There are devices used by radio stations to flip polarity automatically so that signal peaks are maximized in the positive direction. This enables AM stations to transmit at higher power without overmodulation. There are devices that rotate polarity 90 degrees at all frequencies and these completely eliminate any audible polarity. I believe these can improve the "punch" of FM stations. These 90 degree rotators are called "Hilbert Transformers" and are used in some single sideband radio transmitters.
Tony Lauck
"Perception, inference and authority are the valid sources of knowledge" - P.R. Sarkar
This post is made possible by the generous support of people like you and our sponsors:
Follow Ups
- Automatic Detection / Manipulation of Polarity - Tony Lauck 20:14:32 09/24/08 (1)
- RE: Automatic Detection / Manipulation of Polarity - rick_m 21:45:41 09/24/08 (0)