Does anyone know what the numerical Spheric Speech activation settings in Phonak Target actually correspond to acoustically?
For example, if activation is set to 4, 6, 8, 10, 12, etc., can these be translated into approximate ambient noise levels in , or are they based on some other internal measure such as estimated or AutoSense classification?
I’m trying to understand the practical difference between, say, activation at 8 versus 12: under what measured real-world noise conditions should Spheric Speech actually engage at each setting?
Has anyone measured this, or does Phonak publish the relationship anywhere? A simple test with a sound-level meter and controlled background noise would be extremely interesting.
I am not sure but will follow-up with Phonak about it. I suspect it would have more to do with aggressiveness () than ambient noise levels as it is present in manual Spheric Speech in Loud Noise programs.
Not the answer you're seeking however Dr. Cliff states in this video1 that activation of Spheric Speech in Loud Noise for Infinio is 69 db; and Phonak Eon is 66 db. That gives you a default starting point. Standard Speech in Noise activates at 60 db. Is there perhaps a 10 db difference between the default "10" and "0" on the slide rule?
Thanks for the replies. I think the important distinction is between the general Spheric activation threshold and the numerical activation control in Target.
We do have a Phonak representative stating that EON normally moves into Spheric Speech in Loud Noise at roughly 68 . And we know conceptually that lowering the Target activation setting causes Spheric to become eligible sooner.
What we still do not know is what the actual Target numbers mean:
0, 2, 4, 6, 8, 10, 12, etc.
Are they directly related to ? Is one numerical step approximately one ? Is the scale nonlinear? Does enter into it? Is it really an AutoSense probability/classification threshold rather than a simple acoustic level?
Extrapolating from a roughly 68 dB default threshold doesn’t answer that. It still doesn’t tell us what changing 10 to 6, or 6 to 0, actually does acoustically.
And frankly, we shouldn’t have to be having this discussion at all. This is a fitting control that determines when one of Phonak’s most important speech-in-noise technologies comes into play. Yet customers and clinicians are on a hearing forum trying to reverse-engineer what the numbers on the control mean.
It strikes me as a little like putting an important control on a fire extinguisher, labeling it only 0–20, and then telling people merely that “lower makes it activate sooner.” The obvious next question is: so what does each setting actually do?
If the entire purpose of this technology is to help people with hearing loss hear their world better, shouldn’t Phonak tell the professionals fitting it—and ideally the people wearing it—what these settings mean and what changing them is expected to accomplish?
I’d still be very interested in either an official Phonak explanation or an actual controlled measurement of the different settings. Until then, the original question really remains unanswered.
Something I used to do to.lesrn things like this was to make a radical temporary change to one setting. Then wear the aids with this new setting and pay attention to anything that has changed, good or bad. Then go the other direction to see the difference.
This practice was very helpful with many settings that were described as something that didn't make sense.
That makes sense, and I may try exactly that as a way of learning what the control is doing in practice.
What I’m still hoping to pin down, though, is what the numbers themselves represent. A large temporary change can tell me that a setting matters and roughly what direction it pushes the sound, but it still won’t tell me whether the scale represents , an activation threshold, processing strength, a classifier parameter, or something else entirely.
Has anyone seen anything from Phonak — Target documentation, a fitting guide, training material, tooltip, etc. — that actually defines the Spheric Speech Clarity values or explains what changes internally as the number is raised or lowered?
That’s really the missing piece I’m trying to find.
If you look across manufacturers you'll find that most features are hidden under nondescript sliders with numerical values that aren't specifically associated with anything obvious.
Often the sliders are simultaneously controlling multiple parameters and the control has been simplified to include combinations that the manufacturer has found to be useful, excluding those that aren't.
Audiologists have access to manufacturer phone support so we can ask questions about specific clients we have sitting with us to help improve performance, but we don't generally get a full technical rundown of the specifics of how every slider works because we couldn't remember the details for every feature for every product line from every manufacturer even if we wanted to and manufacturers don't want to be totally open about how every feature works to reduce the risk of duplication by competitors.
It's not easy to reverse engineer many sound processing features. Wider input dynamic ranges and Phonak copying Starkey's phase-tagging feedback management system are a couple I can think of over the past 20 years.
Thank you — this is probably the most transparent explanation in the thread, and I appreciate it. My criticism isn’t directed at audiologists or at your explanation. If anything, what you’ve described is exactly what bothers me.
I completely understand why manufacturers would not want to publish proprietary algorithms or enough technical detail for competitors to duplicate them. And I understand that a single slider may actually be controlling several parameters simultaneously.
But there is a big difference between protecting how an algorithm is implemented and explaining what a fitting control is intended to do.
If the Spheric activation scale from 0–20 is not a direct scale, fine — say that. If it is a composite control affecting several internal criteria, say that. If moving it from 10 to 4 causes Spheric to become eligible substantially earlier under certain acoustic conditions, describe that in operational terms and explain the likely tradeoffs.
Nobody is asking Phonak to publish source code or the recipe for Spheric Speech Clarity. I’m asking why a clinician should be presented with a numbered fitting control without being given enough information to know what those numbers actually represent.
The fact that audiologists may need to call manufacturer support about an individual patient because the underlying behavior isn’t adequately documented seems to confirm the problem rather than solve it. And patients certainly shouldn’t have to reverse-engineer the practical meaning of a fitting control on a hearing forum.
Competitive secrecy can reasonably protect how the technology works. I don’t think it should prevent Phonak from explaining how to use the control intelligently.
So your explanation makes a lot of sense to me. It’s the reality you’re describing that I find remarkable.
And patients certainly shouldn’t have to reverse-engineer the practical meaning of a fitting control on a hearing forum.
No, they shouldn't. And this isn't a scenario that any manufacturer would ever consider possible because their software is intended for professional use only, as they make clear in their license agreement for the software. You publicly criticizing them for it isn't good for their reputation. There's nothing stopping them from requiring a professional login to use their software. Count your blessings.
“Count your blessings” cuts both ways. One of mine is that Phonak isn’t the only hearing-aid company in the market. There are other manufacturers with sophisticated -based processing, some offering very competitive performance through lower-cost channels that don’t necessarily require a several-thousand-dollar bundled purchase.
If a manufacturer’s response to reasonable questions about how a fitting control works were to make its software harder to access rather than explain the control better, that would certainly be its prerogative. It would also give consumers another reason to consider competing products.
I’m not asking Phonak to disclose proprietary algorithms. I’m asking for enough information to understand what changing a numbered fitting control actually does. I don’t think that’s an unreasonable expectation for a premium medical device.
Phonak Target is the most end user friendly hearing aid software on the market. It was designed for clinicians rather than engineers and everyone other manufacturer has followed it to improve ease of use. See BHM's BHMFit2 software3 to get an idea of the level of expertise required of clinicians for programming hearing aids in devices with far fewer features than current products.
Phonak's direct relationship is with clinicians. Not end-users. They have an obligation to provide end-users with a product that is reliable and delivers on its promises. Not to create programming software that meets your needs. Your demands aren't reasonable.
im not defending phonak but DIYers are a tiny minority of users. they definitely don't care if we stop buying the hearing aids for the software quirks. most people wearing aids barely know how to use their own phones.
it would be nice to know what particular settings do but they seem to make it opaque for the pros too. is it so their features and algorithms don't get copied? 🤷🏽♀️
I think the fact that DIYers are a tiny minority is beside the point. You’ve actually identified the issue when you say the controls appear opaque to professionals too.
Most hearing-aid users certainly aren’t going to learn Target, nor should they have to. That makes it more important that the clinician fitting them knows what a control does and what tradeoffs come with changing it.
Protecting an algorithm from competitors is understandable. But protecting how something works internally is different from explaining what happens when a clinician changes a control.
A pharmaceutical company doesn’t have to disclose how it manufactures a drug in order to tell physicians what different doses are expected to do. Likewise, Phonak doesn’t need to publish the Spheric algorithm to explain what moving an activation control from 10 to 4 is intended to accomplish.
Otherwise we end up with sophisticated features that people paid for, controls clinicians can adjust, and no clear way for either clinician or patient to know whether those features are being used optimally except by trial and error. That seems like a waste of the technology.
I think this is getting a bit overblown. The slider represents "strength of noise reduction". If everyone had identical hearing loss and identical acoustic coupling maybe that could be represented with a scale but because we don't it can't.
And yet if it represented "strength of noise reduction in dB steps" that would ALSO be a simplification because the is more or less effective at different frequencies and for different types of noises.
I can throw a hearing aid into the testbox and give you some graphs of what it is doing for a particular hearing loss and a closed coupling, but that too will be a simplification because it's a very specific measurement parameter that was developed to test older hearing aids using older noise reduction technologies and it is a very specific noise presentation that isn't that true to life.
You could say, "Okay, I want to know what it is doing in the real world" and develop a cute measurement system whereby people went out into different sound environments and repeated words in noise to see how they did and you could quantify it as number of dB required to get 50% correct and see how that changes with each slider level, but again it's going to be different for each individual who has different hearing loss, different acoustic coupling, and different auditory/cognitive skills.
So, knowing that whatever measure we use is going to be simplified and not precisely applicable to individuals, what sort of measure would you want? I do think that it might be nice to have a standard that we could use to compare hearing aids against one another, although at the same time we do not actually want manufacturers to build their noise reduction systems to optimize a lab-based standard, we want them to build for the real world.
I think you're on target with this response. The effect of the sliders will be different for different people and so a 1 - 20 scale is probably as specific as any control can get. At least there is a number associated with whatever the end-user's preference is so that this can be recorded, alongside non-preferred settings so clinicians can slowly move things towards the preferred end point. Because at the end of the day its all about user preference with these things.
I hope you'll weigh in on a question I provided in another thread related to Phonak devices. I'm on my 3rd set of and approaching 10 years total use. Recognizing speech in noise (restaurants in particular) is my greatest current difficulty. While I leave my HAs at home when I ride my road bike, that doesn't put me at risk because I can hear the low-pitched sounds of approaching traffic reasonably well and just don't LIKE the whistling noise the HAs add to gentle breezes.
And, although I'm a retired physician, the technical details of my hearing deficit elude me; I'm happy to get informed opinions from providers regarding my needs. But I'll suggest that the current MARKETING of solutions to recognition of speech in noisy environments is marketed as "the best thing since sliced bread." If indeed that's the case, I would think that could be easily demonstrated to the potential customer via an in-office comparison of the user's current devices vs. properly adjusted new ones that offer the AI; e.g., listening to a recorded noisy-environment discussion with the new vs. the old HAs, and I'm surprised that's not part of the typical evaluation.
Following this because I'm interested in these kinds of details. It's almost certainly a heuristic based on multiple variables like loudness, time, etc.
@jimrobertson Real world use is going to be a better benchmark than any in-office demonstration especially for ML-assisted speech extraction like Sphere. I've worn hearing aids for a long time and as much as I hate hyping "", I'd say it's a significant development based on my limited experience.
IIRC, you has written you will add your in your profile settings and your % scores for each ear separately and binaurally. Without that, accurate advices is very hard.
us
London
Poland