- KVRAF
- 3501 posts since 7 Sep, 2002

As I wasn't able to obtain this information on the net, and conducting an experiment is a bit out of my reach, I have to ask for an opinion here.

Initially, you have a sound source facing you directly, 12 hours (if you look at it from the top). Then you rotate the sound source relative to yourself clockwise e.g. to 3 hours, or to the right side. If the sound source is a sinewave, for such case do you add phase e.g. cos(x+ph) or subtract phase cos(x-ph)?

I always assumed you add phase when rotating clockwise, but I may be wrong.

Initially, you have a sound source facing you directly, 12 hours (if you look at it from the top). Then you rotate the sound source relative to yourself clockwise e.g. to 3 hours, or to the right side. If the sound source is a sinewave, for such case do you add phase e.g. cos(x+ph) or subtract phase cos(x-ph)?

I always assumed you add phase when rotating clockwise, but I may be wrong.

- KVRist
- 169 posts since 19 Jul, 2008

The phase is related to the time, not the position, so there is no phase shift since the distance to your left ear is kept equal, assuming I understood your question correctly. The amplitude may change a bit, because our ears have a non-omni polar pattern, especially for higher frequencies. (See the patterns measured at https://3diosound.com/products/free-spa ... microphone.) There is also the transfer function of the room to consider.

Last edited by vortico on Mon Jun 25, 2018 1:09 am, edited 1 time in total.

VCV Rack open-source virtual modular synthesizer

- KVRian
- 712 posts since 25 Apr, 2011

Assuming your setup has the sound source rotating, in a horizontal plane, around the centre of your head:

At 12 o'clock the source is directly in front of you, and equidistant from each ear, so both ears will receive the same phase. At 3 o'clock the source is to your right: it is nearer to the right ear and further from the left, so the signal will be advanced in the right and retarded in the left (relative to the 12 o'clock signal) - for the right ear you'd reduce the phase, and for the left ear you'd increase it. You can use geometry to work out the phase difference - the overall time delay is the distance from the sound to the ear divided by the speed of sound, and the phase delay is the time delay multiplied by the frequency.

As vortico mentioned, you will probably also need to adjust volume levels and frequency response for an accurate simulation.

At 12 o'clock the source is directly in front of you, and equidistant from each ear, so both ears will receive the same phase. At 3 o'clock the source is to your right: it is nearer to the right ear and further from the left, so the signal will be advanced in the right and retarded in the left (relative to the 12 o'clock signal) - for the right ear you'd reduce the phase, and for the left ear you'd increase it. You can use geometry to work out the phase difference - the overall time delay is the distance from the sound to the ear divided by the speed of sound, and the phase delay is the time delay multiplied by the frequency.

As vortico mentioned, you will probably also need to adjust volume levels and frequency response for an accurate simulation.

- KVRAF
- 3501 posts since 7 Sep, 2002

vortico wrote:The phase is related to the time, not the position, so there is no phase shift since the distance to your left ear is kept equal, assuming I understood your question correctly. The amplitude may change a bit, because our ears have a non-omni polar pattern, especially for higher frequencies. (See the patterns measured at https://3diosound.com/products/free-spa ... microphone.) There is also the transfer function of the room to consider.

I'm not specifically considering a case of "two ears", it may be a single microphone in front of me. Phase does change - e.g. if you move sound source behind, 180 degrees, phase inverts in practice as far as I know. Why? Because membrane oscillates in opposite direction, for the same distance and sound source. I only have to work out some logic for continuous rotation.

- KVRAF
- 3501 posts since 7 Sep, 2002

kryptonaut wrote:Assuming your setup has the sound source rotating, in a horizontal plane, around the centre of your head:

At 12 o'clock the source is directly in front of you, and equidistant from each ear, so both ears will receive the same phase. At 3 o'clock the source is to your right: it is nearer to the right ear and further from the left, so the signal will be advanced in the right and retarded in the left (relative to the 12 o'clock signal) - for the right ear you'd reduce the phase, and for the left ear you'd increase it. You can use geometry to work out the phase difference - the overall time delay is the distance from the sound to the ear divided by the speed of sound, and the phase delay is the time delay multiplied by the frequency.

As vortico mentioned, you will probably also need to adjust volume levels and frequency response for an accurate simulation.

This logic is reasonable if you only consider time delay factor, in that case phase is reduced. But if you factor in microphone or ear's membrane oscillation, it's not as obvious. Consider the case when it's not 2 ears, but a single microphone and the distance to the sound source does not change.

- KVRAF
- 3501 posts since 7 Sep, 2002

vortico wrote:In the case of your source being a perfectly symmetric vibrating membrane, you will hear nothing when it is rotated 90 degrees, not a phase shift. However, most sources approximate an omnidirectional source.

Let's consider cardioid or omnidirectional mics, not figure-8 mics.

- KVRist
- 169 posts since 19 Jul, 2008

Regardless of the microphone, the membrane will emit no sound in the direction 90 degrees from its "front". What you're describing as a membrane is a figure-8 source. Omnidirectional sources don't invert the signal like you've described above.

VCV Rack open-source virtual modular synthesizer

- KVRAF
- 3501 posts since 7 Sep, 2002

vortico wrote:Regardless of the microphone, the membrane will emit no sound in the direction 90 degrees from its "front". What you're describing as a membrane is a figure-8 source. Omnidirectional sources don't invert the signal like you've described above.

Source is directional. But the mic is fixed in its position. That's where phase relationships arise: relationship between fixed mic and directional sound source.

- KVRian
- 712 posts since 25 Apr, 2011

Maybe it would help if you gave more details of exactly what setup you're trying to model. A sound source would not normally be treated like a double-sided vibrating membrane.

However if you took such a membrane and just rotated it in front of the microphone then there would be an amplitude swing from positive to negative and back to positive again, but no phase change as such (treating the 180 degree change as a negative amplitude). The curve of the amplitude change would depend on the nature of the membrane and what it is supported by.

However if you took such a membrane and just rotated it in front of the microphone then there would be an amplitude swing from positive to negative and back to positive again, but no phase change as such (treating the 180 degree change as a negative amplitude). The curve of the amplitude change would depend on the nature of the membrane and what it is supported by.

- KVRist
- 169 posts since 19 Jul, 2008

We're on the same page. But you are confusing time-phase with angle. When you keep the source and mic at equal distances, there is no phase shift. When you rotate them relative to each other, their transfer functions change, determined by both of their polar responses. Since you have a directional source (say, figure-8 since you said membrane), rotating it relative to the source will to from gain=1 at 0 deg, gain=0 at 90 deg, and gain=-1 at 180 deg. But as kryptonaut said above, figure-8 sources are rare and are probably not what you want unless you have a particular reason for it.

VCV Rack open-source virtual modular synthesizer

- KVRAF
- 3501 posts since 7 Sep, 2002

Well, you are talking about "gain" and "polar pattern", that's not what I'm looking for. In acoustics, where there's gain there's phase incidence. I'm looking for phase incidence between fixed-positioned mic and rotating directional sound source facing such mic directly.

I think you insist that if mic is omnidirectional, the sound source positioned 6 hours won't produce phase inversion in comparison to 12 hours, is this really the case?

I think you insist that if mic is omnidirectional, the sound source positioned 6 hours won't produce phase inversion in comparison to 12 hours, is this really the case?

- KVRist
- 169 posts since 19 Jul, 2008

Of course, if you change the distance between objects, the phase will change. If you read kryptonaut's first post about the problem's geometry, you'd have your question answered. Just calcuate how far the object is from each ear in position A and the distances in position B. Divide by the speed of sound to get the time difference. This time difference gives rise to a phase difference in your original sine wave. However, if you rotate the source with a pivot on your left ear instead of the top of your head, the "source to left ear" transfer function will not be changed, so there is no effect for that ear. Do some trigonometry to derive the relationship between angle and distance, and you'll see that the difference of distances goes from 0 to 10cm when moving from your front to your left.

Last edited by vortico on Mon Jun 25, 2018 6:32 am, edited 2 times in total.

VCV Rack open-source virtual modular synthesizer

- KVRAF
- 3501 posts since 7 Sep, 2002

vortico wrote:Of course, if you change the distance between objects, the phase will change. If you read kryptonaut's first post, you'd have your question answered. Just calcuate how far the object is from each ear in position A and the distances in position B. Divide by the speed of sound to get the time difference. If you focus on the left ear for example, there might be a 5cm change of distance. Do some trigonometry to derive the relationship between angle and distance, and you'll understand the time delays.

I've already replied that the case I'm researching there's no distance difference between the sound source and the recepting (mic) side, only angle changes.