|
Here are some remarks about this feature, motivated by an earlier Twitter conversation and others.
Because “binaural” has been used so many times for completely different technologies, it would be great to have a more specific name for it. My bid would be “Coherent Sideband Binaural” (CSB) – any other suggestions?
I don’t have the time or expertise to write a “manual” but hope this may clarify a few points, maybe clear up some confusion, and help manage unrealistic expectations. Apparently, some are disappointed Binaural does not deliver for them.
Binaural is not stereo. It does not create some dramatic new effect to play with. Its effect is subtle, it can be useful in some situations, and it requires skill using to best advantage, plus headphones. Cloth-ears are a disability in this respect, as I’m discovering with sadness.
There are bound to be flaws in this presentation, not least because of limited expertise and unreliable memory. Also, there is no time available for more experiments, and it is not convenient to run SDR# just now.
I hope others with more (or different) experience will contribute where that helps to correct or expand on the following.
/cont'd
|
Part 1. Background
As it happens, the term “binaural” (two/both ears) pre-dates “stereo” by many decades. Since then, both have been used and misused for a wide range of technologies besides what we (now) call “stereo(phonic)”.
1.1. Stereo(phonic)
Transmission, reproduction and recording of “real” stereo is well over a century old. And (fun fact) one of the earliest demonstrations used quadrature modulations onto a single carrier.
My definition of “real” has the sense of representing a true spatial spread of several sound sources on a single sound stage.
To a good approximation, those sources are "strung out” on a line from left to right – so (in my book) “stereo” qualifies as “1-dimensional” sound, in contrast to “0-dimensional” point sources.
[Stereo “depth” is a concept I have great difficulty with. But, better not to digress.]
I have heard this called “3D”.
Well, if you add the vertical dimension then it might be called “2-D”; and with front/back that might be “3-D” – or how about “surround sound”?
[Whether our ears/brain can truly locate sources unambiguously in 3-D is an interesting question. My suspicion is the brain uses other cues and stored knowledge to “fill in the gaps”.]
The key take-away is that “stereo” attempts to reproduce with fidelity for the listener some experience of an originally spread-out performance.
It is wanton disregard of common usage and common sense to apply the term “stereophonic” to every (or any) other kind of spread audio that is not rooted in spatial fidelity to original material. You might just as well claim that looking through a kaleidoscope is a kind of stereoscopic vision.
In my book, any attempt at producing a “stereo-like” effect from a mono source should be qualified as “pseudo-stereo” or not use the word “stereo” at all.
1.2. Dummy Head
“Dummy Head” capture of a sound stage has its roots in the 1920s but only developed into a serious technology gradually over the next 50 years. The aim was to take account of head and ear geometry for a more “realistic” stereo listener experience.
Dummy head was mostly concerned with stereo recording and has apparently now been tried for surround sound. Certainly, there are demonstrations that claim to provide better height and depth as well as left-right stereo.
Personally, I’m a sceptic. Firstly, because sounds can be engineered for any degree of “effect”. Secondly, because real-world sound engineers for the most “serious” kind of stereo imaging (e.g. classical concerts) still rely on multi-microphone recording and a lot of mixing!
In general, with music and cinema, fidelity plays second fiddle to effect, and so dummy head recording has little industry relevance now.
Confusingly, dummy head recordings are usually called “binaural” (maybe because of the unwanted connotation of “dummy”?).
1.3. Pseudo Stereo
There seems to be no accepted definition and I propose “any technique for creating a field of audio images, similar to stereo, from source audio that has no intrinsic spatial spread” – or “to override/modify any intrinsic spread”.
The motivation for pseudo stereo may be to enhance a listener’s ability to discriminate between sound components. Or it may be purely/largely for effect.
IMO a good example of “useful” pseudo stereo is for CW signals, translating frequency difference into left-right spread (using an audio cross-over for left and right channels).
At the other extreme might be reversing the phase of one channel in a standard stereo audio set-up, creating a weird “phasey” and “disembodied” sound. There are also techniques for widening a stereo image.
Somewhere in between we might include routing SDR I and Q components separately to each ear (more about that later).
With few exceptions, pseudo stereo makes absolutely zero contribution to enhancing intelligibility. Of course, anyone is free to get high on FX, but claims for a more “insightful” experience are suspect until proven.
1.4. AM Sidebands and Demodulation
Amplitude modulation of a carrier creates “sidebands” either side which are mirror images in the perfect case. These are plain to see on any SDR spectrum/FFT, but the received signal will have lost some part of original symmetry because of selective propagation effects.
Recovering the original modulating signal is the function of demodulation. For most of radio history this was performed easily and cheaply with an "envelope" detector. Mathematically, this sums the lower- and upper-sideband content.
Another approach is to filter the incoming signal to isolate a single sideband, but that wastes half the incoming energy.
More advanced methods can use “product” detectors. One such technique uses both sidebands and a lock to the original carrier phase (synchronous AM or SAM demodulation).
SAM has considerable advantages compared with both envelope detection and filtered sideband systems.
Interestingly, sideband selection was first achieved in the late 1940s (at General Electric Company in the USA) using the “phasing method”. This requires a 2-channel signal path with product mixers fed by identical local oscillator except for a differential 90-degree phase shift. Their outputs are what are now well-known to SDR users as the I and Q signals.
There is one more step needed (and this is what doomed the phasing method as soon as Collins started producing ceramic filters for receiver intermediate frequency): one of the I and Q channels must be phase-shifted by a compensating 90 degrees – let us say we transform Q to Q^.
Once that is done, USB = I - Q^, LSB = I + Q^ .... Equation 1
Even today, achieving the extremely tight phase angle tolerance over an entire audio bandwidth is very difficult using analogue circuitry. But digital signal processing (DSP) makes things “easier” in our SDR software (at the expense of some noticeable processing delay).
1.5. Binaural History
GEC’s Donald Norgaard designed their YRS-1 “phasing-method sideband selector” in about 1950. This rather expensive unit tapped into a receiver’s final IF and returned the processed audio which could be one sideband or the other, or both (effectively SAM).
The real thinking man, though, was GEC’s George Floyd who produced a column for their house magazine as “Lighthouse Larry”. In 1951 he adapted the YRS-1 adaptor to route lower and upper sidebands simultaneously to the left and right channels of “split” headphones.
Floyd called YRS-1 a “truly wonderful gimmick” and YRS-1, with his “certain changes”, “this binaural system”. Something of an understatement, IMO.
The binaural presentation technique generated some enthusiasm at the time, but it quickly faded from notice as the phasing method of sideband selection gave way to the easier and cheaper filter method.
However, its memory was kept alive by a few souls, notably (on many occasions) by the UK’s Pat Hawker (G3VA) who also remarked in his February 1989 Technical Topics article for RSGB Magazine that “Spycatcher Peter Wright described how binaural techniques have been used for many years by MI5 phone-tap transcribers to improve the intelligibility of noisy tapes”.
It would take the better part of another half-century before “binaural” became mainstream in SDR software.
1.6. My Interest in Binaural
As a kid, I lapped up amateur radio and short wave listening – especially the esoteric and cutting-edge. Precious sources were my dad’s old Wireless World, RSGB Bulletin, and whatever else pocket money could afford.
I was never really a constructor but spent many sleepless nights on short waves with an ancient HRO, hiding the terrible din inside uncomfortable headphones.
Binaural had become an obsession by the mid-1960s and every resurgence of radio interest turned to disappointment because “no-one had done it yet”. Finally, in retirement with a couple of SDRs, Gnu Radio Companion looked to be a possible way into DSP for dummies like me.
It took at least six months of miserable slog to get a binaural receiver working well enough to convince me it was “real”. Then in November 2020 I had the pleasure of giving a demo to SDRplay staff at a local radio fair.
That has not so far turned into a feature of SDRuno – if it ever does… But I like to kid myself that its appearance in SDR Console six months later was no accident. At least that thought keeps me warm at night.
|
Part 2. Binaural in a Nutshell
Or “Sideband Binaural” (SBB)? Or “Coherent Sideband Binaural” (CSB?)?
2.1. Basics
Just to emphasise (and get this out of the way): binaural is not a demodulation method, it is an audio presentation system for coherently demodulated AM sidebands that are fed to separate audio channels.
Anyone with High School mathematics should be able to understand the maths behind Norgaard’s sideband selector (and Floyd’s binaural). All you need are a few trigonometric identities – which about sums up my level of (in)expertise.
It takes rather more effort to understand how the resultant audio is affected by de-tuning, and a lot more to see what sideband imbalance from propagation can do. And engineers who design SDR software to implement binaural are in a class of their own.
However, there is an intuitive way to see what we should expect.
Anyone having some familiarity with quadrature demodulators will realise that the I-channel delivers just the AM modulation, the Q-channel just the FM (or phase) modulation.
If we rearrange the previous Equation 1, we get (ignoring amplitude):
I = LSB + USB, Q = (LSB - USB)^ .... Equation 2
And those who have listened to I & Q channels separately for an AM station will be aware that I-channel sounds roughly “normal” while Q is “almost nothing”.
So, for AM (DSB with carrier), I-channel is the averaged amplitude modulation content and Q has (a phase-shifted version of) the frequency modulation content – normally zero, but essentially the difference between sidebands during disturbances.
From an information point of view, AM DSB is a monaural source. And it is heard as such in good reception conditions.
Binaural recovers any sideband imbalance to present as a kind of “difference” signal that spreads the audio image by the degree of that imbalance.
This is a little reminiscent of conventional stereo broadcasting where the main signal is the average soundstage, and the difference signal is the left-right imbalance – but contrasting, in that ideally there is no spread with binaural.
In principle, binaural works well on Independent Sideband (ISB) signals too (given at least a residual carrier), except that the listener may be more confused than when listening to them separately.
There may be little benefit in applying binaural to single-sideband signals, but I have no experience yet in this area.
And it should be obvious that binaural will be essentially deaf to frequency/phase modulated signals (Q has the modulation, with the I-channel essentially nothing).
2.2. Unpacked
Coherent Sideband Binaural is what I have described above, using product detectors and the phasing method. It features:
· A two-channel baseband created from a source split into two mixers, with local oscillator feeds in relative quadrature (90-degree phase difference), giving output channels conventionally designated I (in-phase) and Q (quadrature). So far, so standard SDR architecture.
· A compensating quadrature relative shift applied between I and Q.
· Sum and difference circuits to isolate LSB and USB.
· Crucially, the signal paths must be phase-coherent (so, for example, just taking a filtered LSB from one virtual receiver and USB from another is a non-starter).
Its operation is mathematically precise and uncontroversial. There is no room for “customising” or “improving”. It is what it is.
NO OTHER ARRANGEMENT works (correctly) like this, whether called “binaural” or not. Rick Campbell’s design in the March 1999 QST, that some erroneously claim to be a precedent, merely feeds I and Q signals to left and right ’phones.
I-Q to L-R does not deliver the benefits of CSB. There is (at first hearing) some resemblance and some educational value. But that is the best that can be said about it. Mathematically, it is total nonsense and should detain us no longer – despite its possible attraction for pseudo stereo fans.
With the few exceptions below, all other so-called “binaural” systems (e.g. FlexRadio’s) either present I & Q components as raw data to the left and right audio channels (as Campbell) or are based on some kind of pseudo stereo algorithm.
“True” CSB was implemented by:
· Simon Brown in SDR Console (2021)
· “jks” in KiwiSDR (a little later, and called “SAS”)
· Youssef Touil in SDR# (2022)
At a slightly earlier date, one German developer wrote a similar feature in software for a group project based on USRP (or something similar) that never really got off the ground. But, as far as I know after extensive research, no other publicly available software ever had this capability.
|
Part 3. Binaural Listening
First impressions can be very confusing for those not already clued-up, so here are a few pointers.
3.1. It’s Not Stereo!
If you engage Binaural plus Carrier Lock and listen to a strong station with little adjacent channel activity, what you should get is very stable audio in mono! Each sideband is a good copy of the other, so almost identical signals appear in both audio channels. Perfect balance equals centre stage.
In other words, binaural does not “create a stereo effect” – that is not its purpose.
What you will notice, if there is a little background noise, is that the noise is not sitting directly “on top” of the wanted signal (as it is with AM or SAM). Rather, it is spread evenly across the sound stage. Because noise is uncorrelated with the desired signal, there is an effective improvement in signal-to-noise ratio.
That alone probably accounts for an IEEE evaluation over 60 years ago that binaural makes at least a 3 dB intelligibility improvement even over synchronous AM (and attracted MI5).
3.2. Noisy Neighbours
Keeping the same settings, choose another strong station but with a strong signal on an adjacent channel. Hopefully, you will find that the neighbour sounds mainly “off to one side” where he belongs – not right on top of the wanted station, as with other modes.
In general, it should be possible to “tame” sideband splatter and other kinds of off-channel interference without aggressive narrowing of the IF passband.
3.3. Fading/Propagation
Skywave propagation is typically dogged by selective fading, each sideband affected differently. With standard AM demodulators, the sound may suffer from the “Luxembourg effect” (for those old enough to remember).
Synchronous demodulation offers better immunity to selective fading, largely removing distortion, but binaural presents this in a quite different fashion: as the sidebands’ phase relationship to the carrier varies, we hear a kind of rolling “phasiness”. This may not necessarily be good for intelligibility, nor for a weak stomach. But it does what it does :).
3.4. Locked and Unlocked Carrier
As described before, with a “perfect” signal, absent propagation effects (e.g. with a local station), a correctly-tuned and phase-locked binaural receiver will produce a 100% mono audio image, exactly centred (if your ears are good), just as intended.
SDR# tunes the LO itself so you get an accurate reading of the station’s actual transmitted frequency – subject to receiver calibration, naturally, and to a visual precision of 1 Hz (though behind the scenes it is better).
Disabling lock without disturbing the formerly locked LO frequency, you may hear little or no difference to the audio. More likely, you will hear a slow “pumping” as the relative phase of received and reinserted carriers cycles at a rate somewhat below 1 Hz.
The “pumping” is accompanied by a degree of “phasiness” as the two sidebands alternate in relative phase.
Now, if you tune (unlocked) further away from the transmitted frequency, by up to (say) 20 Hz, the “beat” frequency rises accordingly and is heard as rapid “flutter”. This also begins to destroy the harmonic balance of the audio, producing the well-known “mis-tuned SSB” effect (though with a suppressed carrier the flutter is absent).
You may also notice a further effect – frequency-dependent “echo” off to the sides (a nuisance side-effect but not serious).
When the degree of mistuning exceeds about 60 Hz, a beat tone will also appear. There should also be a noticeable displacement of the audio image to one side of centre.
3.5. “Width” of the Binaural Soundstage
Questions for future research, not covered in my early experiments:
· At what deviation from centre frequency will an adjacent station sound fully “side-lined”?
· Is there a roughly linear relationship between deviation and “azimuth”?
Answers will give us a yardstick to estimate frequency deviation from aural displacement. A supplementary question is whether there is scope for adjusting the soundstage width (and whether that could be beneficial).
3.6. PLL Characteristics
It is vital, for binaural operation, and for synchronous demodulation in general, that the carrier reinsertion signal is locked to the carrier of the signal of interest. The technique is a phase-locked loop (PLL).
PLL design is tricky, having to work well for a variety of reception conditions and to suit the listener’s needs. The SDR software designer may provide fixed characteristics or choose to allow user adjustment of certain parameters. That is well outside my comfort zone, and I don’t intend to get involved.
However, what the listener will notice is behaviour of the lock operation (when engaged). If you select binaural lock while tuned remote from a station, there will be no lock. When tuning towards a station, there comes an offset at which the PLL locks – that illustrates the loop capture range. Then, tuning away again, lock will be lost beyond the lock range.
To make best use of binaural in difficult situations, it is important to have a good feel for these two parameters. There will be times when varying signal conditions cause loss of lock or for the PLL to re-lock to a different carrier.
The question “to lock or not to lock?” will come up later.
|
Part 4. Crowded Channels
There is no doubt in my mind that binaural really comes into its own when several stations occupy the same channel or overlap. That is common on long/medium and short-wave bands.
MW can be a special case, with different channel allocation across regions, and is famous with transatlantic DX hunters who may have the luxury of finding carrier offsets of 1, 2, or 3 kHz from local stations.
Then there are many other cases of significant deviation from nominal frequency, often detectable in the spectrum or waterfall as having a carrier distinct from “well-behaved” occupants. And (anyway), with a high-resolution FFT, you may spot many carriers “off” by something under 50 Hz.
Trying to resolve any of these may be assisted by binaural – sometimes carrier-locked, but (perhaps more often) unlocked.
4.1. Transatlantic DX
This is mainly a winter-time activity and most successful in the early hours of the morning (IQ recordings to the rescue). New entrants to the fun may find it easier to start with “graveyard” channels with little or no local-area channel occupants.
Once a good candidate has been spotted significantly offset from the “local” channel’s nominal frequency, there is a reasonable chance of hearing it with binaural locked. In my limited experience, it can be a revelation to have this distant station centred and stable while any neighbouring nuisance is largely banished to the soundstage margins (or otherwise disembodied and unfocused).
Here is a good opportunity to mention a quite different (also revolutionary) tool provided in SDR#: the channel canceller. I have not been able to experiment myself but have enjoyed various video demonstrations that leave me in no doubt about its potential.
There will often be times when the channel canceller is better able to pull a weaker signal out from behind a dominant station.
It remains to be seen whether there is any chance of binaural operating in conjunction with the canceller. Any feedback will be appreciated.
Another point worth mentioning here is the likelihood of carrier heterodynes with intrusive audio at 1, 2 or 3 kHz. A narrow IF or AF notch filter can make a huge difference.
If transatlantic conditions are fair, it is common to hear more than one station – that gets interesting because they may be of comparable strength and vie with each other for carrier lock. This is the cue to disengage binaural lock and try very small tuning steps to see whether each station can be “focused” at the expense of the other. More in the next section.
4.2. Small Carrier Offsets
Whether transatlantic or regional, on low/medium or short-waves, stations with small offsets from nominal are very common. There is a “MW Offsets” user group for enthusiasts and those who want to learn more about this.
Such offsets can be diagnostic for station identification, so there can be a powerful incentive for dedicated DX-ers to explore this subject in more detail.
While some stations are well off nominal (by even hundreds of Hz) and easily recognised on spectrum/waterfall, many deviate by under a few tens of Hz. These can sometimes be spotted by cranking up the resolution bandwidth and zooming the waterfall.
Then a huge number of stations, perhaps the majority, deviate by fractions of 1 Hz. Detecting them, resolving them separately, and identifying their origin push technology and operator skill to their limits – I’ll not have much more to say about these.
My main point on the offsets theme is that binaural sometimes has the potential to give prominence and clarity to one or other station in a group (with roughly comparable received power levels) that otherwise come through as a cacophony. Carrier lock must be disengaged to get results here.
If we have no visual cue to channel occupancy, slowly stepping through tuning either side of nominal may succeed in focusing/stabilising an individual station well enough to achieve identification. If there are favourable carrier offsets, and a high enough waterfall resolution, homing in on each may be easier – more on this later.
4.3. Sub-Herz Tuning
Earlier, I spoke about displayed tuning precision in the case of binaural carrier lock. SDR software commonly has a read-out and setting precision of 1 Hz and, while amazing compared with most receivers a decade ago, is not quite good enough for some specialised work.
Please note that I am not talking about accuracy that involves calibration and frequency stability. It is a question of being able to vary tuning in smaller steps and observe that in the display.
Binaural-locked frequency display has already been mentioned – a read-out to the nearest 1 Hz does not prepare you for the slow “pumping” that may then occur after disengaging the lock. Neither does an unlocked 1 Hz tuning step always allow you to “tame” the slowly changing phase error of any individual station.
Of course, a transmitter may not be quite stable in frequency, and anyway your receiver may be drifting, so there is no guarantee of long-term “perfection”.
While milli-Herz precision tuning/display may be the goal for ultra-specialists, I do wonder if a compromise improvement may be possible in SDR#.
|
[Edit in 5.1]
Part 5. Further Investigations and Possibilities
This is a very brief list, for future reference and to stimulate discussion.
5.1. Compatibility with Other SDR# Tools
I questioned earlier whether binaural might work in conjunction with:
· co-channel canceller
It would be interesting to hear about compatibility/benefits with other tools:
· audio/IF notch(es)
· noise reduction [Edit: really happy to hear from prog this plays nicely]
· IF passband filters (symmetric and otherwise)
Suggestions for other combinations are welcome.
5.2. Higher-Resolution Tuning
As described in the last part, sub-Herz frequency read-out could be very helpful.
I have played with tuning steps down to 1 mHz (thousandth) and have convinced myself smaller steps are useful. However, I found it very difficult to know when the optimum had been passed – a tuning aid is needed. A rough-and-ready I-Q plot with rotating spot helped a little, but there could be better displays?
5.3. Higher-Resolution Waterfall
Visual recognition and estimation of small carrier offsets could be assisted by an enhanced zoom capability. There is, of course a trade-off between resolution and update speed – the question is what might reasonably be possible.
/ENDOF
|
Just FWIW,
About 10 years ago I programmed a dual channel NetSDR (not many were made) so that Antenna 1 routed to the left audio, Antenna 2 to the right, both channels having identical DSP. The brain did the heavy lifting, the result was quite amazing.
Not if only prog would make a 2 or 4 channel phase coherent 0 to 2 GHz SDR receiver for $199.
переключити цитоване повідомлення
Показати цитований текст
From: airspy@groups.io <airspy@groups.io> on behalf of Paul White via groups.io <paul@...>
Sent: 27 July 2022 01:34
To: airspy@groups.io <airspy@groups.io>
Subject: [airspy] Binaural Discovery
Here are some remarks about this feature, motivated by an earlier Twitter conversation and others.
Because “binaural” has been used so many times for completely different technologies, it would be great to have a more specific name for it. My bid would be “Coherent Sideband Binaural” (CSB) – any other suggestions?
I don’t have the time or expertise to write a “manual” but hope this may clarify a few points, maybe clear up some confusion, and help manage unrealistic expectations. Apparently, some are disappointed Binaural does not deliver for them.
Binaural is not stereo. It does not create some dramatic new effect to play with. Its effect is
subtle, it can be useful in some situations, and it requires
skill using to best advantage, plus headphones. Cloth-ears are a disability in this respect, as I’m discovering with sadness.
There are bound to be flaws in this presentation, not least because of limited expertise and unreliable memory. Also, there is no time available for more experiments, and it is not convenient to run SDR# just
now.
I hope others with more (or different) experience will contribute where that helps to correct or expand on the following.
/cont'd
|
CW "brain selectivity" in stereo field.
Listen to CW in a wider bandwidth in DSB binaural, and you will find all stations within the bandwidth "sit" on their own position in the stereo field, depending on tone offset from the center.
Signals are nicely separable. A good op could even decode multiple stations by ear.
I found this to be a fascinating addition in CW pile ups.
73
Paul
PD0SWL
|
Hi Paul, A fascinating and informative read for me, many thanks. 73 John.
[Edited Message Follows]
CW "brain selectivity" in stereo field.
Listen to CW in a wider bandwidth in DSB binaural, and you will find all stations within the bandwidth "sit" on their own position in the stereo field, depending on tone offset from the center.
Signals are nicely separable. A good op could even decode multiple stations by ear.
I found this to be a fascinating addition in CW pile ups.
73
Paul
PD0SWL
|
Independent Sideband, ISB, survived the mechanical resonator
filter (not ceramic) and crystal filter but not in the ham realm.
I might still have advertisements from the 60s for them. I know I
do not have the mil spec to which I worked at one point that
called for ISB. The data on the two sidebands was not required to
be related. It did require a modest pilot carrier to work.
Your interest involves a subset of ISB. It's something I kept
running across very infrequently over the years through about 1980
when I moved more into software.
And to me "binaural" means different information to the two ears
usually related to each other but not always. "Stereo" defines a
specific relationship. And my misdirected comment about the I and
Q to different ears article had brought to mind pseudo-stereo that
leveraged on an artifact of conventional musical instrument
placement. It sent higher frequencies to one ear and lower to the
other with overlap more or less encouraged.
I *think* from this word blizzard you are speaking of ISB with
one sideband being left and the other right. It is a compatible
means of sending genuine if possibly "attenuated separation"
stereo over an AM channel. My aging memory more or less insists
that much of the experimentation with it was based in the Midwest.
(But it does not insist that very loudly.)
Feeding I and Q to different ears gives your ears a dimension
that can be used to separate signals that fall through normal
filters using human type audio processing. I suspect a person who
gets used to it may find it quite useful. Other users may find the
pseudo-stereo presentation useful as it moves signals within the
filter bandwidth from ear to ear when using wider filters, which
are nicer for very fast CW. (My NC109's crystal filter could notch
or peak a chunk of spectrum. The very narrow chunk setting caused
enough ringing copying CW was painful. But, then, it always was
for me. The ringing just made it worse.)
Noting I/Q reception with some signals, DSB for example, you can
lock to the missing carrier with a conventional phase lock loop, a
Costas loop. If the sidebands differ you need a real if perhaps
suppressed carrier for locking purposes.
How close did I get to the thrust of your thinking? For what you
are describing the reference must be sent with the information or
it will not demodulate correctly. It is a nice technique to
possibly resolve some kinds if path distortion that affects one
sideband and not the other. Adding them back together should be
done in the head rather fed to a speaker.
I'd hesitate about calling int binaural. Too many other
applications have used the term. So it's not a good name from a
standpoint of distinguishing it from other uses of "Binaural" that
use the dual related but different audio processing paths in the
usual human head.
{^_^}
On 20220726 21:51:54, Paul White wrote:
переключити цитоване повідомлення
Показати цитований текст
Part 2.
Binaural in a Nutshell
Or “Sideband
Binaural” (SBB)? Or “Coherent Sideband Binaural” (CSB?)?
2.1.
Basics
Just to emphasise
(and get this out of the way): binaural is not a
demodulation method, it is an audio presentation
system for coherently demodulated AM sidebands
that are fed to separate audio channels.
Anyone with High
School mathematics should be able to understand the maths behind
Norgaard’s sideband selector (and Floyd’s binaural). All you
need are a few trigonometric identities – which about sums up my
level of (in)expertise.
It takes rather
more effort to understand how the resultant audio is affected by
de-tuning, and a lot more to see what sideband imbalance from
propagation can do. And engineers who design SDR software to
implement binaural are in a class of their own.
However, there is
an intuitive way to see what we should expect.
Anyone having
some familiarity with quadrature demodulators will realise that
the I-channel delivers just the AM modulation, the Q-channel
just the FM (or phase) modulation.
If we rearrange
the previous Equation 1, we get (ignoring amplitude):
I = LSB + USB, Q
= (LSB - USB)^ ....
Equation 2
And those who
have listened to I & Q channels separately for an AM station
will be aware that I-channel sounds roughly “normal” while Q is
“almost nothing”.
So, for AM (DSB
with carrier), I-channel is the averaged amplitude modulation
content and Q has (a phase-shifted version of) the frequency
modulation content – normally zero, but essentially the
difference between sidebands during disturbances.
From an information
point of view, AM DSB is a monaural source. And it is
heard as such in good reception conditions.
Binaural recovers
any sideband imbalance to present as a kind of
“difference” signal that spreads the audio image by
the degree of that imbalance.
This is a
little reminiscent of conventional stereo broadcasting
where the main signal is the average soundstage, and the
difference signal is the left-right imbalance – but contrasting,
in that ideally there is no spread with binaural.
In principle,
binaural works well on Independent Sideband (ISB) signals too
(given at least a residual carrier), except that the listener
may be more confused than when listening to them separately.
There may be
little benefit in applying binaural to single-sideband signals,
but I have no experience yet in this area.
And it should be
obvious that binaural will be essentially deaf to
frequency/phase modulated signals (Q has the modulation, with
the I-channel essentially nothing).
2.2.
Unpacked
Coherent Sideband
Binaural is what I have described above, using product detectors
and the phasing method. It features:
· A
two-channel baseband created from a source split into two
mixers, with local oscillator feeds in relative quadrature
(90-degree phase difference), giving output channels
conventionally designated I (in-phase) and Q (quadrature). So
far, so standard SDR architecture.
· A
compensating quadrature relative shift applied between I and Q.
· Sum
and difference circuits to isolate LSB and USB.
· Crucially,
the signal paths must be phase-coherent (so, for
example, just taking a filtered LSB from one virtual receiver
and USB from another is a non-starter).
Its operation is
mathematically precise and uncontroversial. There is no room for
“customising” or “improving”. It is what it is.
NO OTHER
ARRANGEMENT works (correctly) like this, whether called
“binaural” or not. Rick Campbell’s design in the March 1999 QST,
that some erroneously claim to be a precedent, merely feeds I
and Q signals to left and right ’phones.
I-Q to L-R does not
deliver the benefits of CSB. There is (at first hearing) some
resemblance and some educational value. But that is the best
that can be said about it. Mathematically, it is total nonsense
and should detain us no longer – despite its possible attraction
for pseudo stereo fans.
With the few
exceptions below, all other so-called “binaural”
systems (e.g. FlexRadio’s) either present I & Q components
as raw data to the left and right audio channels (as Campbell)
or are based on some kind of pseudo stereo algorithm.
“True” CSB was
implemented by:
· Simon
Brown in SDR Console (2021)
· “jks”
in KiwiSDR (a little later, and called “SAS”)
· Youssef
Touil in SDR# (2022)
At a slightly
earlier date, one German developer wrote a similar feature in
software for a group project based on USRP (or something
similar) that never really got off the ground. But, as far as I
know after extensive research, no other publicly available
software ever had this capability.
|
In practical terms now that I have equipment capable of sub-Hz
resolution it is more or less mot very usable on HF. The
ionosphere happily introduces frequency offsets as it moves up and
down. On VHF and further up it might help if the GD oscillators
were really really stable. Otherwise you calmly sit and watch the
oscillator drift as the AC kicks in and out. Nonetheless, it can
be done, easily, and might well be worth doing at the digital
level. To avoid annoying other users I'd make it a configuration
item.
Yes, when sitting around of an evening nattering with the group
around W6OBB late at night I did sit there watching phase both
wibble around and go off to a high rate of change. Many in the
group tried to send "manly" signals with lots of low frequencies.
So tuning plus or minus 0.5 Hz or so (whatever the ProII actually
supplied) was nice to have. When tuned that close the duck-talk
sound went away, which is the usual objection to SSB from the
diehard AM folks.
So I am not overly convinced of the utility of futility of adding
sub-Hz tuning. But, again, as an obscure configuration item why
not go for it. Maybe Paul is onto something bigger than I see
here.
{o.o}
On 20220726 21:56:26, Paul White wrote:
переключити цитоване повідомлення
Показати цитований текст
Part 5.
Further Investigations and Possibilities
This is a very
brief list, for future reference and to stimulate discussion.
5.1.
Compatibility with Other SDR# Tools
I questioned
earlier whether binaural might work in conjunction with:
· co-channel
canceller
It would be
interesting to hear about compatibility/benefits with other
tools:
· audio/IF
notch(es)
· noise
reduction
· IF
passband filters (symmetric and otherwise)
Suggestions for
other combinations are welcome.
5.2.
Higher-Resolution Tuning
As described in
the last part, sub-Herz frequency read-out could be very
helpful.
I have played
with tuning steps down to 1 mHz (thousandth) and have convinced
myself smaller steps are useful. However, I found it very
difficult to know when the optimum had been passed – a tuning
aid is needed. A rough-and-ready I-Q plot with rotating spot
helped a little, but there could be better displays?
5.3.
Higher-Resolution Waterfall
Visual
recognition and estimation of small carrier offsets could be
assisted by an enhanced zoom capability. There is, of course a
trade-off between resolution and update speed – the question is
what might reasonably be possible.
/ENDOF
|
Um for that price - it's to dream about. I wonder if $499 would
sell such a thing in suitable quantities.
{O.O}
On 20220726 22:53:31, Simon Brown
wrote:
переключити цитоване повідомлення
Показати цитований текст
Just FWIW,
About 10 years ago I programmed a dual channel NetSDR (not many
were made) so that Antenna 1 routed to the left audio, Antenna 2
to the right, both channels having identical DSP. The brain did
the heavy lifting, the result was quite amazing.
Not if only prog would make a 2 or 4 channel phase coherent 0 to
2 GHz SDR receiver for $199.
Here are some
remarks about this feature, motivated by an earlier Twitter
conversation and others.
Because
“binaural” has been used so many times for completely
different technologies, it would be great to have a more
specific name for it. My bid would be “Coherent Sideband
Binaural” (CSB) – any other suggestions?
I don’t have the time or expertise to write a “manual” but
hope this may clarify a few points, maybe clear up some
confusion, and help manage unrealistic expectations.
Apparently, some are disappointed Binaural does not deliver
for them.
Binaural
is not stereo. It does not create some dramatic new
effect to play with. Its effect is
subtle, it can be useful in some
situations, and it requires
skill using to best advantage, plus headphones.
Cloth-ears are a disability in this respect, as I’m
discovering with sadness.
There are
bound to be flaws in this presentation, not least because of
limited expertise and unreliable memory. Also, there is no
time available for more experiments, and it is not convenient
to run SDR# just now.
I hope others
with more (or different) experience will contribute where that
helps to correct or expand on the following.
/cont'd
|
Agreed. Spatial diversity with separated antennas can give an
amazing listening experience! Especially on CW and SSB. I also
experimented with this stuff years ago but with 2-channel direct
conversion hardware with a common LO. I got inspired by another
article by Rick Campbell KK7B titled "Binaural presentation of SSB
and CW signals received on a pair of antennas". IIRC that PDF was
included on the CD that came with the EMRFD book.
@prog: Thanks for the ISB addition to SDR#! I adds another
dimension to listening, literally.
Johan
Simon Brown wrote:
переключити цитоване повідомлення
Показати цитований текст
Just FWIW,
About 10 years ago I programmed a dual channel NetSDR (not many
were made) so that Antenna 1 routed to the left audio, Antenna 2
to the right, both channels having identical DSP. The brain did
the heavy lifting, the result was quite amazing.
Not if only prog would make a 2 or 4 channel phase coherent 0 to
2 GHz SDR receiver for $199.
|
Hi jdow, Thank you.
In all of the groups I am in, your posts are always an education and inspiration for me. You have knowledge far beyond anything that I will ever have and most everything you post is always an educational journey and experience for me.
I thank you for your knowledge and everything else that you share. Best Wishes and Best Regards.
73. John
On Wed, 27 Jul 2022 at 08:55, jdow < jdow@...> wrote:
Independent Sideband, ISB, survived the mechanical resonator
filter (not ceramic) and crystal filter but not in the ham realm.
I might still have advertisements from the 60s for them. I know I
do not have the mil spec to which I worked at one point that
called for ISB. The data on the two sidebands was not required to
be related. It did require a modest pilot carrier to work.
Your interest involves a subset of ISB. It's something I kept
running across very infrequently over the years through about 1980
when I moved more into software.
And to me "binaural" means different information to the two ears
usually related to each other but not always. "Stereo" defines a
specific relationship. And my misdirected comment about the I and
Q to different ears article had brought to mind pseudo-stereo that
leveraged on an artifact of conventional musical instrument
placement. It sent higher frequencies to one ear and lower to the
other with overlap more or less encouraged.
I *think* from this word blizzard you are speaking of ISB with
one sideband being left and the other right. It is a compatible
means of sending genuine if possibly "attenuated separation"
stereo over an AM channel. My aging memory more or less insists
that much of the experimentation with it was based in the Midwest.
(But it does not insist that very loudly.)
Feeding I and Q to different ears gives your ears a dimension
that can be used to separate signals that fall through normal
filters using human type audio processing. I suspect a person who
gets used to it may find it quite useful. Other users may find the
pseudo-stereo presentation useful as it moves signals within the
filter bandwidth from ear to ear when using wider filters, which
are nicer for very fast CW. (My NC109's crystal filter could notch
or peak a chunk of spectrum. The very narrow chunk setting caused
enough ringing copying CW was painful. But, then, it always was
for me. The ringing just made it worse.)
Noting I/Q reception with some signals, DSB for example, you can
lock to the missing carrier with a conventional phase lock loop, a
Costas loop. If the sidebands differ you need a real if perhaps
suppressed carrier for locking purposes.
How close did I get to the thrust of your thinking? For what you
are describing the reference must be sent with the information or
it will not demodulate correctly. It is a nice technique to
possibly resolve some kinds if path distortion that affects one
sideband and not the other. Adding them back together should be
done in the head rather fed to a speaker.
I'd hesitate about calling int binaural. Too many other
applications have used the term. So it's not a good name from a
standpoint of distinguishing it from other uses of "Binaural" that
use the dual related but different audio processing paths in the
usual human head.
{^_^}
On 20220726 21:51:54, Paul White wrote:
Part 2.
Binaural in a Nutshell
Or “Sideband
Binaural” (SBB)? Or “Coherent Sideband Binaural” (CSB?)?
2.1.
Basics
Just to emphasise
(and get this out of the way): binaural is not a
demodulation method, it is an audio presentation
system for coherently demodulated AM sidebands
that are fed to separate audio channels.
Anyone with High
School mathematics should be able to understand the maths behind
Norgaard’s sideband selector (and Floyd’s binaural). All you
need are a few trigonometric identities – which about sums up my
level of (in)expertise.
It takes rather
more effort to understand how the resultant audio is affected by
de-tuning, and a lot more to see what sideband imbalance from
propagation can do. And engineers who design SDR software to
implement binaural are in a class of their own.
However, there is
an intuitive way to see what we should expect.
Anyone having
some familiarity with quadrature demodulators will realise that
the I-channel delivers just the AM modulation, the Q-channel
just the FM (or phase) modulation.
If we rearrange
the previous Equation 1, we get (ignoring amplitude):
I = LSB + USB, Q
= (LSB - USB)^ ....
Equation 2
And those who
have listened to I & Q channels separately for an AM station
will be aware that I-channel sounds roughly “normal” while Q is
“almost nothing”.
So, for AM (DSB
with carrier), I-channel is the averaged amplitude modulation
content and Q has (a phase-shifted version of) the frequency
modulation content – normally zero, but essentially the
difference between sidebands during disturbances.
From an information
point of view, AM DSB is a monaural source. And it is
heard as such in good reception conditions.
Binaural recovers
any sideband imbalance to present as a kind of
“difference” signal that spreads the audio image by
the degree of that imbalance.
This is a
little reminiscent of conventional stereo broadcasting
where the main signal is the average soundstage, and the
difference signal is the left-right imbalance – but contrasting,
in that ideally there is no spread with binaural.
In principle,
binaural works well on Independent Sideband (ISB) signals too
(given at least a residual carrier), except that the listener
may be more confused than when listening to them separately.
There may be
little benefit in applying binaural to single-sideband signals,
but I have no experience yet in this area.
And it should be
obvious that binaural will be essentially deaf to
frequency/phase modulated signals (Q has the modulation, with
the I-channel essentially nothing).
2.2.
Unpacked
Coherent Sideband
Binaural is what I have described above, using product detectors
and the phasing method. It features:
· A
two-channel baseband created from a source split into two
mixers, with local oscillator feeds in relative quadrature
(90-degree phase difference), giving output channels
conventionally designated I (in-phase) and Q (quadrature). So
far, so standard SDR architecture.
· A
compensating quadrature relative shift applied between I and Q.
· Sum
and difference circuits to isolate LSB and USB.
· Crucially,
the signal paths must be phase-coherent (so, for
example, just taking a filtered LSB from one virtual receiver
and USB from another is a non-starter).
Its operation is
mathematically precise and uncontroversial. There is no room for
“customising” or “improving”. It is what it is.
NO OTHER
ARRANGEMENT works (correctly) like this, whether called
“binaural” or not. Rick Campbell’s design in the March 1999 QST,
that some erroneously claim to be a precedent, merely feeds I
and Q signals to left and right ’phones.
I-Q to L-R does not
deliver the benefits of CSB. There is (at first hearing) some
resemblance and some educational value. But that is the best
that can be said about it. Mathematically, it is total nonsense
and should detain us no longer – despite its possible attraction
for pseudo stereo fans.
With the few
exceptions below, all other so-called “binaural”
systems (e.g. FlexRadio’s) either present I & Q components
as raw data to the left and right audio channels (as Campbell)
or are based on some kind of pseudo stereo algorithm.
“True” CSB was
implemented by:
· Simon
Brown in SDR Console (2021)
· “jks”
in KiwiSDR (a little later, and called “SAS”)
· Youssef
Touil in SDR# (2022)
At a slightly
earlier date, one German developer wrote a similar feature in
software for a group project based on USRP (or something
similar) that never really got off the ground. But, as far as I
know after extensive research, no other publicly available
software ever had this capability.
|
Radio and then software were basically my life - my monomania du
jour. Of late I've become astonished at how much other stuff I
crammed into my head without being aware of it. Things like
history are starting to make sense to me. Perhaps that is my age
and reactions like, "Gee, I saw this exact same silliness in
1967." (And now I doubly wish POTUSs Carter and Clinton had never
existed. And Woke is just plain stupid. I saw how life was for
blacks in the 50s. I see how it is now. People are complaining as
if it was still the 50s. They have nothing to complain about today
by comparison. The perceive problem was being fixed organically.
Now they are amplifying the problem. I guess the WOKEites want the
blacks pushed down into feeling useless and persecuted as a vote
farm.) But, this does not belong here. Seeing this, though, is an
artifact of simple being there at the time with a mind like a
sponge. Squeeze it a little and only God knows what will come out.
{o.-}
On 20220727 01:15:40, John Bain wrote:
переключити цитоване повідомлення
Показати цитований текст
Hi jdow, Thank you.
In all of the groups I am in, your posts are
always an education and inspiration for me. You have knowledge
far beyond anything that I will ever have and most everything
you post is always an educational journey and experience for
me.
I thank you for your knowledge and everything
else that you share.
Best Wishes and Best Regards.
73. John
On Wed, 27 Jul 2022 at
08:55, jdow < jdow@...>
wrote:
Independent Sideband, ISB, survived the mechanical
resonator filter (not ceramic) and crystal filter but
not in the ham realm. I might still have advertisements
from the 60s for them. I know I do not have the mil spec
to which I worked at one point that called for ISB. The
data on the two sidebands was not required to be
related. It did require a modest pilot carrier to work.
Your interest involves a subset of ISB. It's something
I kept running across very infrequently over the years
through about 1980 when I moved more into software.
And to me "binaural" means different information to the
two ears usually related to each other but not always.
"Stereo" defines a specific relationship. And my
misdirected comment about the I and Q to different ears
article had brought to mind pseudo-stereo that leveraged
on an artifact of conventional musical instrument
placement. It sent higher frequencies to one ear and
lower to the other with overlap more or less encouraged.
I *think* from this word blizzard you are speaking of
ISB with one sideband being left and the other right. It
is a compatible means of sending genuine if possibly
"attenuated separation" stereo over an AM channel. My
aging memory more or less insists that much of the
experimentation with it was based in the Midwest. (But
it does not insist that very loudly.)
Feeding I and Q to different ears gives your ears a
dimension that can be used to separate signals that fall
through normal filters using human type audio
processing. I suspect a person who gets used to it may
find it quite useful. Other users may find the
pseudo-stereo presentation useful as it moves signals
within the filter bandwidth from ear to ear when using
wider filters, which are nicer for very fast CW. (My
NC109's crystal filter could notch or peak a chunk of
spectrum. The very narrow chunk setting caused enough
ringing copying CW was painful. But, then, it always was
for me. The ringing just made it worse.)
Noting I/Q reception with some signals, DSB for
example, you can lock to the missing carrier with a
conventional phase lock loop, a Costas loop. If the
sidebands differ you need a real if perhaps suppressed
carrier for locking purposes.
How close did I get to the thrust of your thinking? For
what you are describing the reference must be sent with
the information or it will not demodulate correctly. It
is a nice technique to possibly resolve some kinds if
path distortion that affects one sideband and not the
other. Adding them back together should be done in the
head rather fed to a speaker.
I'd hesitate about calling int binaural. Too many other
applications have used the term. So it's not a good name
from a standpoint of distinguishing it from other uses
of "Binaural" that use the dual related but different
audio processing paths in the usual human head.
{^_^}
On 20220726 21:51:54, Paul White wrote:
Part 2. Binaural in a
Nutshell
Or “Sideband Binaural” (SBB)? Or
“Coherent Sideband Binaural” (CSB?)?
2.1. Basics
Just to emphasise (and get this
out of the way): binaural is not a
demodulation method, it is an audio
presentation system for coherently
demodulated AM sidebands that are fed to
separate audio channels.
Anyone with High School
mathematics should be able to understand the maths
behind Norgaard’s sideband selector (and Floyd’s
binaural). All you need are a few trigonometric
identities – which about sums up my level of
(in)expertise.
It takes rather more effort to
understand how the resultant audio is affected by
de-tuning, and a lot more to see what sideband
imbalance from propagation can do. And engineers who
design SDR software to implement binaural are in a
class of their own.
However, there is an intuitive way
to see what we should expect.
Anyone having some familiarity
with quadrature demodulators will realise that the
I-channel delivers just the AM modulation, the
Q-channel just the FM (or phase) modulation.
If we rearrange the previous
Equation 1, we get (ignoring amplitude):
I = LSB + USB, Q = (LSB - USB)^
.... Equation 2
And those who have listened to I
& Q channels separately for an AM station will be
aware that I-channel sounds roughly “normal” while Q
is “almost nothing”.
So, for AM (DSB with carrier),
I-channel is the averaged amplitude modulation content
and Q has (a phase-shifted version of) the frequency
modulation content – normally zero, but essentially
the difference between sidebands during disturbances.
From an information
point of view, AM DSB is a monaural source.
And it is heard as such in good reception conditions.
Binaural recovers any sideband imbalance
to present as a kind of “difference” signal that spreads
the audio image by the degree of that
imbalance.
This is a little
reminiscent of conventional stereo broadcasting where
the main signal is the average soundstage, and the
difference signal is the left-right imbalance – but
contrasting, in that ideally there is no spread
with binaural.
In principle, binaural works well
on Independent Sideband (ISB) signals too (given at
least a residual carrier), except that the listener
may be more confused than when listening to them
separately.
There may be little benefit in
applying binaural to single-sideband signals, but I
have no experience yet in this area.
And it should be obvious that
binaural will be essentially deaf to frequency/phase
modulated signals (Q has the modulation, with the
I-channel essentially nothing).
2.2. Unpacked
Coherent Sideband Binaural is what
I have described above, using product detectors and
the phasing method. It features:
· A
two-channel baseband created from a source split into
two mixers, with local oscillator feeds in relative
quadrature (90-degree phase difference), giving output
channels conventionally designated I (in-phase) and Q
(quadrature). So far, so standard SDR architecture.
· A
compensating quadrature relative shift applied between
I and Q.
· Sum
and difference circuits to isolate LSB and USB.
· Crucially,
the signal paths must be phase-coherent (so,
for example, just taking a filtered LSB from one
virtual receiver and USB from another is a
non-starter).
Its operation is mathematically
precise and uncontroversial. There is no room for
“customising” or “improving”. It is what it is.
NO OTHER ARRANGEMENT works
(correctly) like this, whether called “binaural” or
not. Rick Campbell’s design in the March 1999 QST,
that some erroneously claim to be a precedent, merely
feeds I and Q signals to left and right ’phones.
I-Q to L-R does not
deliver the benefits of CSB. There is (at first
hearing) some resemblance and some educational value.
But that is the best that can be said about it.
Mathematically, it is total nonsense and should detain
us no longer – despite its possible attraction for
pseudo stereo fans.
With the few exceptions below, all
other so-called “binaural” systems (e.g. FlexRadio’s)
either present I & Q components as raw data to the
left and right audio channels (as Campbell) or are
based on some kind of pseudo stereo algorithm.
“True” CSB was implemented by:
· Simon
Brown in SDR Console (2021)
· “jks”
in KiwiSDR (a little later, and called “SAS”)
· Youssef
Touil in SDR# (2022)
At a slightly earlier date, one
German developer wrote a similar feature in software
for a group project based on USRP (or something
similar) that never really got off the ground. But, as
far as I know after extensive research, no other
publicly available software ever had this capability.
--
John Bain
|
No politics here, please. jdow <jdow@...> skrev: (27 juli 2022 10:25:59 CEST)
Radio and then software were basically my life - my monomania du
jour. Of late I've become astonished at how much other stuff I
crammed into my head without being aware of it. Things like
history are starting to make sense to me. Perhaps that is my age
and reactions like, "Gee, I saw this exact same silliness in
1967." (And now I doubly wish POTUSs Carter and Clinton had never
existed. And Woke is just plain stupid. I saw how life was for
blacks in the 50s. I see how it is now. People are complaining as
if it was still the 50s. They have nothing to complain about today
by comparison. The perceive problem was being fixed organically.
Now they are amplifying the problem. I guess the WOKEites want the
blacks pushed down into feeling useless and persecuted as a vote
farm.) But, this does not belong here. Seeing this, though, is an
artifact of simple being there at the time with a mind like a
sponge. Squeeze it a little and only God knows what will come out.
{o.-}
On 20220727 01:15:40, John Bain wrote:
Hi jdow, Thank you.
In all of the groups I am in, your posts are
always an education and inspiration for me. You have knowledge
far beyond anything that I will ever have and most everything
you post is always an educational journey and experience for
me.
I thank you for your knowledge and everything
else that you share.
Best Wishes and Best Regards.
73. John
On Wed, 27 Jul 2022 at
08:55, jdow < jdow@...>
wrote:
Independent Sideband, ISB, survived the mechanical
resonator filter (not ceramic) and crystal filter but
not in the ham realm. I might still have advertisements
from the 60s for them. I know I do not have the mil spec
to which I worked at one point that called for ISB. The
data on the two sidebands was not required to be
related. It did require a modest pilot carrier to work.
Your interest involves a subset of ISB. It's something
I kept running across very infrequently over the years
through about 1980 when I moved more into software.
And to me "binaural" means different information to the
two ears usually related to each other but not always.
"Stereo" defines a specific relationship. And my
misdirected comment about the I and Q to different ears
article had brought to mind pseudo-stereo that leveraged
on an artifact of conventional musical instrument
placement. It sent higher frequencies to one ear and
lower to the other with overlap more or less encouraged.
I *think* from this word blizzard you are speaking of
ISB with one sideband being left and the other right. It
is a compatible means of sending genuine if possibly
"attenuated separation" stereo over an AM channel. My
aging memory more or less insists that much of the
experimentation with it was based in the Midwest. (But
it does not insist that very loudly.)
Feeding I and Q to different ears gives your ears a
dimension that can be used to separate signals that fall
through normal filters using human type audio
processing. I suspect a person who gets used to it may
find it quite useful. Other users may find the
pseudo-stereo presentation useful as it moves signals
within the filter bandwidth from ear to ear when using
wider filters, which are nicer for very fast CW. (My
NC109's crystal filter could notch or peak a chunk of
spectrum. The very narrow chunk setting caused enough
ringing copying CW was painful. But, then, it always was
for me. The ringing just made it worse.)
Noting I/Q reception with some signals, DSB for
example, you can lock to the missing carrier with a
conventional phase lock loop, a Costas loop. If the
sidebands differ you need a real if perhaps suppressed
carrier for locking purposes.
How close did I get to the thrust of your thinking? For
what you are describing the reference must be sent with
the information or it will not demodulate correctly. It
is a nice technique to possibly resolve some kinds if
path distortion that affects one sideband and not the
other. Adding them back together should be done in the
head rather fed to a speaker.
I'd hesitate about calling int binaural. Too many other
applications have used the term. So it's not a good name
from a standpoint of distinguishing it from other uses
of "Binaural" that use the dual related but different
audio processing paths in the usual human head.
{^_^}
On 20220726 21:51:54, Paul White wrote:
Part 2. Binaural in a
Nutshell
Or “Sideband Binaural” (SBB)? Or
“Coherent Sideband Binaural” (CSB?)?
2.1. Basics
Just to emphasise (and get this
out of the way): binaural is not a
demodulation method, it is an audio
presentation system for coherently
demodulated AM sidebands that are fed to
separate audio channels.
Anyone with High School
mathematics should be able to understand the maths
behind Norgaard’s sideband selector (and Floyd’s
binaural). All you need are a few trigonometric
identities – which about sums up my level of
(in)expertise.
It takes rather more effort to
understand how the resultant audio is affected by
de-tuning, and a lot more to see what sideband
imbalance from propagation can do. And engineers who
design SDR software to implement binaural are in a
class of their own.
However, there is an intuitive way
to see what we should expect.
Anyone having some familiarity
with quadrature demodulators will realise that the
I-channel delivers just the AM modulation, the
Q-channel just the FM (or phase) modulation.
If we rearrange the previous
Equation 1, we get (ignoring amplitude):
I = LSB + USB, Q = (LSB - USB)^
.... Equation 2
And those who have listened to I
& Q channels separately for an AM station will be
aware that I-channel sounds roughly “normal” while Q
is “almost nothing”.
So, for AM (DSB with carrier),
I-channel is the averaged amplitude modulation content
and Q has (a phase-shifted version of) the frequency
modulation content – normally zero, but essentially
the difference between sidebands during disturbances.
From an information
point of view, AM DSB is a monaural source.
And it is heard as such in good reception conditions.
Binaural recovers any sideband imbalance
to present as a kind of “difference” signal that spreads
the audio image by the degree of that
imbalance.
This is a little
reminiscent of conventional stereo broadcasting where
the main signal is the average soundstage, and the
difference signal is the left-right imbalance – but
contrasting, in that ideally there is no spread
with binaural.
In principle, binaural works well
on Independent Sideband (ISB) signals too (given at
least a residual carrier), except that the listener
may be more confused than when listening to them
separately.
There may be little benefit in
applying binaural to single-sideband signals, but I
have no experience yet in this area.
And it should be obvious that
binaural will be essentially deaf to frequency/phase
modulated signals (Q has the modulation, with the
I-channel essentially nothing).
2.2. Unpacked
Coherent Sideband Binaural is what
I have described above, using product detectors and
the phasing method. It features:
· A
two-channel baseband created from a source split into
two mixers, with local oscillator feeds in relative
quadrature (90-degree phase difference), giving output
channels conventionally designated I (in-phase) and Q
(quadrature). So far, so standard SDR architecture.
· A
compensating quadrature relative shift applied between
I and Q.
· Sum
and difference circuits to isolate LSB and USB.
· Crucially,
the signal paths must be phase-coherent (so,
for example, just taking a filtered LSB from one
virtual receiver and USB from another is a
non-starter).
Its operation is mathematically
precise and uncontroversial. There is no room for
“customising” or “improving”. It is what it is.
NO OTHER ARRANGEMENT works
(correctly) like this, whether called “binaural” or
not. Rick Campbell’s design in the March 1999 QST,
that some erroneously claim to be a precedent, merely
feeds I and Q signals to left and right ’phones.
I-Q to L-R does not
deliver the benefits of CSB. There is (at first
hearing) some resemblance and some educational value.
But that is the best that can be said about it.
Mathematically, it is total nonsense and should detain
us no longer – despite its possible attraction for
pseudo stereo fans.
With the few exceptions below, all
other so-called “binaural” systems (e.g. FlexRadio’s)
either present I & Q components as raw data to the
left and right audio channels (as Campbell) or are
based on some kind of pseudo stereo algorithm.
“True” CSB was implemented by:
· Simon
Brown in SDR Console (2021)
· “jks”
in KiwiSDR (a little later, and called “SAS”)
· Youssef
Touil in SDR# (2022)
At a slightly earlier date, one
German developer wrote a similar feature in software
for a group project based on USRP (or something
similar) that never really got off the ground. But, as
far as I know after extensive research, no other
publicly available software ever had this capability.
--
John Bain
|
The new Perseus22 is ~2100 to 2500 euro, so sadly probably not.
переключити цитоване повідомлення
Показати цитований текст
From: airspy@groups.io <airspy@groups.io> on behalf of jdow via groups.io <jdow@...>
Sent: 27 July 2022 09:10
To: airspy@groups.io <airspy@groups.io>
Subject: Re: [airspy] Binaural Discovery
Um for that price - it's to dream about. I wonder if $499 would sell such a thing in suitable quantities.
{O.O}
On 20220726 22:53:31, Simon Brown wrote:
Just FWIW,
About 10 years ago I programmed a dual channel NetSDR (not many were made) so that Antenna 1 routed to the left audio, Antenna 2 to the right, both channels having identical DSP. The brain did the heavy lifting, the result was quite amazing.
Not if only prog would make a 2 or 4 channel phase coherent 0 to 2 GHz SDR receiver for $199.
Here are some remarks about this feature, motivated by an earlier Twitter conversation and others.
Because “binaural” has been used so many times for completely different technologies, it would be great to have a more specific name for it. My bid would be “Coherent Sideband Binaural” (CSB) – any other suggestions?
I don’t have the time or expertise to write a “manual” but hope this may clarify a few points, maybe clear up some confusion, and help manage unrealistic expectations. Apparently, some are disappointed Binaural does not deliver for them.
Binaural is not stereo. It does not create some dramatic new effect to play with. Its effect is
subtle, it can be useful in some situations, and it requires
skill using to best advantage, plus headphones. Cloth-ears are a disability in this respect, as I’m discovering with sadness.
There are bound to be flaws in this presentation, not least because of limited expertise and unreliable memory. Also, there is no time available for more experiments, and it is not convenient to run SDR# just
now.
I hope others with more (or different) experience will contribute where that helps to correct or expand on the following.
/cont'd
|
Sorry,
Unclear reply by me: $499 is possibly an unrealistic price.
переключити цитоване повідомлення
Показати цитований текст
From: airspy@groups.io <airspy@groups.io> on behalf of Simon Brown via groups.io <simon@...>
Sent: 27 July 2022 12:31
To: airspy@groups.io <airspy@groups.io>
Subject: Re: [airspy] Binaural Discovery
The new Perseus22 is ~2100 to 2500 euro, so sadly probably not.
From: airspy@groups.io <airspy@groups.io> on behalf of jdow via groups.io <jdow@...>
Sent: 27 July 2022 09:10
To: airspy@groups.io <airspy@groups.io>
Subject: Re: [airspy] Binaural Discovery
Um for that price - it's to dream about. I wonder if $499 would sell such a thing in suitable quantities.
{O.O}
On 20220726 22:53:31, Simon Brown wrote:
Just FWIW,
About 10 years ago I programmed a dual channel NetSDR (not many were made) so that Antenna 1 routed to the left audio, Antenna 2 to the right, both channels having identical DSP. The brain did the heavy lifting, the result was quite amazing.
Not if only prog would make a 2 or 4 channel phase coherent 0 to 2 GHz SDR receiver for $199.
Here are some remarks about this feature, motivated by an earlier Twitter conversation and others.
Because “binaural” has been used so many times for completely different technologies, it would be great to have a more specific name for it. My bid would be “Coherent Sideband Binaural” (CSB) – any other
suggestions?
I don’t have the time or expertise to write a “manual” but hope this may clarify a few points, maybe clear up some confusion, and help manage unrealistic expectations. Apparently, some are disappointed Binaural does not deliver for them.
Binaural is not stereo. It does not create some dramatic new effect to play with. Its effect is
subtle, it can be useful in some situations, and it requires
skill using to best advantage, plus headphones. Cloth-ears are a disability in this respect, as I’m discovering with sadness.
There are bound to be flaws in this presentation, not least because of limited expertise and unreliable memory. Also, there is no time available for more experiments, and it is not convenient to run SDR#
just now.
I hope others with more (or different) experience will contribute where that helps to correct or expand on the following.
/cont'd
|
About 10 years ago I programmed a dual channel NetSDR (not many were made) so that Antenna 1 routed to the left audio, Antenna 2 to the right, both channels having identical DSP. The brain did the heavy lifting, the result was quite amazing.
"DX-er's Dream", Simon :). When it comes to identification, what better than a parallel channel? Even LW ~ VHF, or MW ~ internet radio, or validating a jingle against your own audio snippet library. And they don't even have to be coherent. Just once, I made the effort with Console to feed one station to a simple audio mixer, and a web browser with web radio version to another input. Balanced the audio and had a magic experience. Totally and badly out of sync, but totally successful. As you say, the brain does the heavy lifting. It would be nice to have a stand-alone software audio mixer desk. Something like an easy-to-use, drastically cut-down VoiceMeeter. 6-channel VAC input, punch up any into L, any into R, balance/panning control. And, SDR outputs as audio end-points (not just "available" for VAC), user-defined label for station/frequency. Quick-record of single (time-stamped) "it's not stereo" for library/QSL. CSV User List Browser et al making it easier to tabulate all candidate parallel broadcast channels. "Automatically" opening an SDR VRX for any one, on request? IQ recording was a revolution, but now we hardly have enough time in a day to digest it properly. And every new DSP trick like channel canceller only makes things "worse". Happy Days.
|
On Wed, Jul 27, 2022 at 08:55 AM, jdow wrote:
Noting I/Q reception with some signals, DSB for example, you can lock to the missing carrier with a conventional phase lock loop, a Costas loop. If the sidebands differ you need a real if perhaps suppressed carrier for locking purposes.
How close did I get to the thrust of your thinking? For what you are describing the reference must be sent with the information or it will not demodulate correctly. It is a nice technique to possibly resolve some kinds if path distortion that affects one sideband and not the other. Adding them back together should be done in the head rather fed to a speaker.
I'd hesitate about calling int binaural. Too many other applications have used the term. So it's not a good name from a standpoint of distinguishing it from other uses of "Binaural" that use the dual related but different audio processing paths in the usual human head.
Yes, Joanne, I did use a Costas Loop in my demo, and wasn't interested in non-DSB. And my "thinking" is entirely in the context of headphone presentation of a single station, pure Lighthouse Larry. That's precisely what Simon/Youssef/Kiwi deliver, and what this thread is all about. Whatever anyone says, "stereo" is now consumer jargon with a very specific meaning. "Binaural" is still, in principle, a generic term. But loaded with too many connotations. We have to call this something different and recognisable, in an attempt to shut down the interminable arguments and crazy distractions. Hence the "[coherent?] sideband binaural" suggestion.
|
Simon Brown
John