• The forum software that supports hummy.tv has been upgraded to XenForo 2.3!

    Please bear with us as we continue to tweak things, and feel free to post any questions, issues or suggestions in the upgrade thread.

Mono from Stereo

I'm sure I have also done the same in the past, and this is the 'naive' method - I put the word in quotes because it turns out that the naive approach is also the considered method (and indeed the only viable method).

As the linked Wikipedia article shows, for any method of making the recording except binaural or time-of-arrival stereophony, there is essentially a monophonic signal available encoded as L+R/2, so the mixing of the left and right channels is entirely justified. In the case of the exceptions, there is no satisfactory way to extract a mono signal (but a minority of recordings will have been made this way).

I am intrigued by the mid+side stereo microphone, I had wondered how a useful stereo separation could be obtained from what appears to be a single mic on (say) a camcorder.
 
The thing that intrigued me was the "false stereo" created by adjusting volumes at the two speakers, rather than phase differences, as we actually recognize directional information.
 
Do we register phase differences? The article implies we register differences in time of arrival - which is not the same thing.

I can see how 'false stereo' (as you put it) could be a proxy for delta-ToA - there will be a distinct and opposite delta for sounds specifically emanating from the L or R speaker, so maybe there is a linear correlation when the sources are varied in relative intensity.
 
Interesting point. Plus, phase differences will vary with frequency. I thought we did register phase at low frequencies but not at higher ones. Sorry, I was not referring solely to that article.

Your OP is relevant to converting a stereo signal to a surround signal, too. What do you send to the centre channel? Are phase differences used in surround processing? Must read more!

Have you got the equipment to send a pure tone via headphones to your left ear and an opposite phase one to your right ear? That would tell you if we can register phase differences!
 
Download the wavs in #5 and #14 ! ! ! I can convert them to mp3 if that is easier, the problem then is the mp3 encoding itself may do strange things
 
If MP3 buggers about with phases and yet we don't notice, that would be a strong indication that we don't get any auditory information from phase directly (only amplitude variation as a result of constructive or destructive interference with a reference signal).

I think the experiment would be better conducted using a sampled real-life waveform, which is then put on the time-line for the left and right headphone channels in a variety of ways:

A. L delayed from R by -0.5ms, -0.4ms... +0.5ms

B. L delayed from -R by -0.5ms, -0.4ms... +0.5ms

Using the same amplitude both sides eliminates any directional cues that would be obtained from one side being louder than the other.

If there is no perceived difference between A and B, we can conclude that phase has nothing to do with it and it's all ToA. It is hard to see how phase could have anything to do with it - the ears operate independently and send spectrum data to the brain without phase info as far as I know, and there is no built-in interferometer.

Another interesting inquiry: when you have, say, a sawtooth wave, does the sound change if the phases of the harmonics are altered relative to the fundamental?

I can do all this, regrettably I don't have time at the moment (and shouldn't even be on here!).
 
If you can't detect phase shift with a single sine wave, I think it's very unlikely you will detect it with a complex waveform
 
Anyway, inter-aural time delay only applies to a sound onset or a short sharp sound. If you are listening to a sine wave, there is no way you can tell what time delay there is. All you get is a phase difference between the ears. That will vary as you turn your head.
 
If you can't detect phase shift with a single sine wave, I think it's very unlikely you will detect it with a complex waveform
Why? I agree we probably don't sense phase information per se, but I don't see how you derive your second clause from the first. It's more likely the other way around. Ever tried locating the source of a continuous pure tone?

I thought you might oblige by creating the test series I specified, if you're going to wait for me to do it you'll be waiting until Friday.
 
Surely ToA and phase are basically the same thing as far as our hearing is concerned. If the sound arrives at one ear later than the other then obviously it will be slightly out of phase - with low notes only a tiny proportion of the wavelength; a bigger portion as frequency rises. Hence we can put a sub-woofer almost anywhere.

But I'm quite sure our (and other animals') hearing is far more sophisticated than just detecting phase/ToA. In simple stereo terms we shouldn't be able to tell if a sound is in front or behind, or indeed above or below, but we can, and without even thinking.
"the ears operate independently and send spectrum data to the brain without phase info as far as I know, and there is no built-in interferometer". You reckon? I'd be amazed if our ears/brains aren't way more capable than that. We have complex flappy things round our ears which must have evolved for a reason, presumably to subtly alter the sound entering the ear from different directions, and our brains can detect and interpret that 'data' in an instant.
 
"Airline pilots exploit the 3 dB [improvement in hearing a signal above background noise] by feeding their receiver's audio out of phase to each ear to suppress cockpit noise."

http://www.nitehawk.com/rasmit/br_cpy.html
I'm not sure what that proves. I can well imagine that by creating two sound channels which are noise+signal and noise+antisignal, at any moment one or other of them will have a better signal to noise ratio and that the brain is good at focusing on the best signal at the time. This would have nothing to do with an ability to perform a correlation analysis on the input from two ears.
 
Why? I agree we probably don't sense phase information per se, but I don't see how you derive your second clause from the first. It's more likely the other way around. Ever tried locating the source of a continuous pure tone?
I thought you might oblige by creating the test series I specified, if you're going to wait for me to do it you'll be waiting until Friday.
With a constant 1kHz tone it is possible to have left and right channels constantly 180 degrees apart, if you replace the tone with a complex waveform containing many frequencies and delay one channel by a fixed time of say 0.5ms you will only get 180 deg phase shift at a single frequency which may never be present in the complex waveform and if it is present, you will only achieve 180 phase shift for the fraction of a second that the correct frequency is present
 
Hi, Ezra

Perhaps 180 degrees apart is misleading. You can take any waveform to the left ear and invert it in the right. The sum of the two signals is zero, but I am sure you would hear both, though it would sound strange. Here is an example, attached: two files, one with both tracks identical, the other with one inverted.
 

Attachments

When I did it a while (long while) ago. I played a 1kHz in the L channel and an anti-phase 1kHz in the other. It did not sound weird at all. It sounded like two 1kHz tones, one coming from the Left and the other coming from, surprise, surprise, the right speaker. In phase it sounds like a 1kHz tone from the centre point of the two speakers just like it should.
 
Perhaps 180 degrees apart is misleading. You can take any waveform to the left ear and invert it in the right. The sum of the two signals is zero, but I am sure you would hear both, though it would sound strange. Here is an example, attached: two files, one with both tracks identical, the other with one inverted.
Agreed, it is possible to invert a complex waveform to so that the entire right hand channel is 180 deg out of phase with the left hand channel, the bit that is wrong in #26, is delaying one of the channels by a fixed amount of time (say 0.5ms) as this will only produce 180 phase shift at a single frequency, leaving all other frequencies shifted by some other amount rather that 180 degrees, if the resulting stereo waveform is then converted to mono, only the very small parts of the original that were exactly 180 degrees out of phase will be set to zero, leaving all other parts of the waveform still being present in the mono version
 
I don't agree. The point of the proposed experiment in post 26 is to see whether the phase change affects perception at all, and is not intended to be listened to from speakers or to test mono extraction. It's crude, but will determine whether the human ear-brain system is sensitive to phase or not: if the first series produces a stereo image shifting across the field, then we will know that ToA is sufficient and phase is not necessary; if the second series also produces the same effect we will know that the stereo effect doesn't give a damn about phase.

I see no point in condemning the experiment before trying it - either wait for me to generate the test signal or generate the specified signal yourself.
 
I don't understand what we are trying to achieve here. Post #26 is ambiguous: L delayed from -R? Do you mean invert L and R, or just R?

"when you have, say, a sawtooth wave, does the sound change if the phases of the harmonics are altered relative to the fundamental?"


The sawtooth has a unique Fourier transform. If you start adding phase differences to the terms in the Fourier series, you won't in general get a sawtooth any more, and it will sound different. Hence, the ear is detecting phase differences between the frequencies it extracts.
 
Back
Top