Well, Ralph was.
Phantom = fake. Is the goal not to sound natural?
Well a majority of those do not cause what we are talking about here. What I am talking about is explained with basic summation of vectors, natural sound or phantom sound in space.
Yes, fake but not necessarily wrong or distinguishable from a real sound source.
Don't know how we ended up here but you claimed that "Stereo requires a specific ammount of crosstalk to work." and "If you don't like the sound trapped inside your head with headphones you need crossfeed to re-introduce an ammount of crosstalk to bring the sound outside of your head with stereo recordings.". Both is not correct. The latter "just" requires the presence of your individual head-related transfer function (HRTF). It can be applied while recording (binaural recording) or while playback (e.g. Smyth Realiser).
Hello,
Stereo (triangle) indeed relies on cross-talk. Amplitude panning transforms in the interference field into ITD cues. This was how Blumlein explained stereo 80 years ago. But the stereo interference field is not natural, the velocity vector is not correct! That's one of the reasons when you turn your head even a little bit the sound image collapses.
- Elias
Stereo (triangle) indeed relies on cross-talk. Amplitude panning transforms in the interference field into ITD cues. This was how Blumlein explained stereo 80 years ago. But the stereo interference field is not natural, the velocity vector is not correct! That's one of the reasons when you turn your head even a little bit the sound image collapses.
- Elias
I agree!
How do you explain localisation of a phantom source of let's say 400Hz amplitude panned signal in the above stereo?
- Elias
By simple interchannel level and/or time differences. But quite honestly, we don't know how summing localisation really works. See http://www.hauptmikrofon.de/theile/ON_THE_LOCALISATION_english.pdf
To make it clearer what crosstalk is:
L(R) and R(L) in b) is crosstalk.
To make it clearer what crosstalk is:
An externally hosted image should be here but it was not working when we last tested it.
L(R) and R(L) in b) is crosstalk.
There is no level difference at low freqs. How do you get time difference from amplitude panned signal?
Only way to get ITD in this case is from cross-talk. This is my point.
- Elias
This is irrelevant and does not qualify the stereo definition we agreed above.
In stereo triangle there is no amplitude difference (in ear signals) for 400Hz amplitude panned signal.
Only with headphones, or when cross talk is removed, you can generate low frequency ILD. Stereo triangle cannot do this! It relies on ITD generated by the cross-talk. That's why I'm saying stereo relies on cross talk.
- Elias
You guys in Finland are pretty "creative" lately
There's no difference in intensity between the left and the right ear when only the right speaker is playing?
You guys in Finland are pretty "creative" lately
Define "difference".
- Elias
There is no level difference at low freqs. How do you get time difference from amplitude panned signal?
Only way to get ITD in this case is from cross-talk. This is my point.
- Elias
Normally humans localize using both interaural level differences and interaural time differences in the frequency range mostly below 1000 Hz. In the Blumlein method of recording, there is no ITD captured since the two figure eight microphones are coincident. So, to make a silk purse out of a sows ear, one can observe (a la Lipschitz) that in the case where the input is not all center, or all side, (so there is sound at different levels coming from each speaker) that at one ear the louder signal is combined with a slightly delayed lower signal and at the other ear the lower signal is combined with a slightly delayed louder signal. If you compare these combined signals at each ear there is now a phase difference between the ears, that would not be there but for the crosstalk. A phase difference can be thought of as an ITD but at subwoofer frequencies this phase shift is on the order of 7 degrees and there is no evidence that this is audible. (What is audible is the rise in bass level for center sounds compared to far side ones.)
As the frequency rises and this crosstalk induced phase difference becomes appreciable one can say there is an audible ITD. But with further rise in frequency the phase difference between the ears reverses polarity and then this form of not-recorded ITD declines.
Of course, this phase shift or induced ITD does not necessarily correspond to the actual localization (ITD) of anything at the recording location. Also the amount of ITD generated this way not only depends on the frequency involved, but also on the exact difference in the level of the channels. Again, as above, starting at say 500 Hz the phase shift across the head will reach 180 degrees so you will just have a lower level at each ear without any phase difference. Again, at very low frequencies the brain cannot detect small phase shifts caused by the distance between the ears so the crosstalk has no effect in the subwoofer region. At all frequencies, the 30 degree placement of the speakers induces a fixed ITD of about 220 microseconds for every signal heard, and this applies both to Blumlein or spaced recordings. How all this unnatural ITD effects each listener I can't tell you. But things sound a lot better without it.
Ralph Glasgal
Home Page
Thanks Ralph. That's why I said "Stereo does not "rely" on crosstalk, it works despite of crosstalk."
I have to admit my listening experience with Ambiophonics is limited but all stereo recordings I tried did sound unnatural to me. Maybe it's because these recordings were made for the equilateral stereo triangle and not for crosstalk corrected playback.
Best, Markus
I have to admit my listening experience with Ambiophonics is limited but all stereo recordings I tried did sound unnatural to me. Maybe it's because these recordings were made for the equilateral stereo triangle and not for crosstalk corrected playback.
Best, Markus
HRTF are an end result, there are many things that comprise HRTF's and crosstalk or if you want to call it vectors instead, are important. You need a vector to determine what HRTF to use.
I like this statement, however in past discussions with others they seem to get very upset and state it "requires" or "relies" on it. I much rather prefer "despite", so no disagreement here.
It was a necessary evil of having no other option at the time.Panpot is not really "Stereo" although it is label as pan-pot stereo. That is the problem, the word "stereo" can mean so many things and is very vague. I consider it a marketing term as Blumlein never called his concept and inventions "stereo". Just consider all the different types of recording in stereo and you have a mess.
By panning sounds in headphones, it is not real and there are problems. This does not happen in nature. You cannot be deaf to sounds coming from the sides in one ear (unless it is high frequencies), it sounds unnatural. This is the reason for simulating HRTF or adding crossfeed, whatever you want to call it.
Last edited: