Phonak Infinio Sphere vs Oticon Intent – quantitative word recognition testing

@DaleM1 said:

This is a 31% word recognition improvement of the Sphere over the Intent. I believe this is statistically significant and corroborates my subjective impressions.

This is consistent with Phonak’s statements. In their evidence whitepaper they write about similar percentage.

@DaleM1, I have a question: can the manual Spheric Speech Clarity programme denoise the sample played by the speakers in a similar way to Adobe Podcast AI (free account)?

Original file with jackhammer:

Deleted2

Sharing files with file.io is convenient, anonymous and secure. Just upload files and share the links via email, SMS, Slack, Discord, etc. Easy-to-use REST API.

sample audio files for speech recognition (kaggle.com)12

After denoising by Adobe Podcast AI:

Deleted

Sharing files with file.io is convenient, anonymous and secure. Just upload files and share the links via email, SMS, Slack, Discord, etc. Easy-to-use REST API.

I’m sorry, I coudn’t find longer samples. Some of it in the Target software:

I’ll give this a try tomorrow. But I’ve read that the spherical speech in loud noise doesn’t work well with a recorded track in which the speech and noise are combined. I believe it works much better when there is spatial separation, which I used for my experiment.

Nice experiment Dale, thanks for sharing.

@Bimodal_user said:

I have a question: can the manual Spheric Speech Clarity programme denoise the sample played by the speakers in a similar way to Adobe Podcast AI (free account)?

I downloaded the files and did the following tests. Files were played back on Bowers&Wilkens PX7 S2 noise cancelling headphones:

1. Original file played back with Phonak “music” program (no compression or other processing except equalization).
2. Original file played back with Phonak “spheric speed in loud noise” program
3. Adobe Podcase AI file played back with Phonak “music” program.
   Results: In all three, the speech is intelligible, except that, to my ear, the word “beer” sounds more like “veer” (all three). S/N level goes from lowest to highest in fairly equal steps. In other words, the “spheric” mode in the Phonaks is a definite improvement, but not as good as Adobe’s (presumably off-line processing). Not bad for real time (6 ms delay).

Referring to your audiologist as your “PEANUT AUDI” doesn’t make people think what you think it does…

Wow, that was creative…

A son has the latest Sony’s and when my regular hearing aids were being repaired he lent me his as he has an earlier Sony pair and we programmed them via their app on my iphone. As someone said there is nothing to it so buying them in a parking lot doesn’t make any difference in programming but might with warranty but I am guessing. They were actually very good and you can see from my audiogram that I have severe hearing loss.

Yeah, I shoulda just said “the IMPATIENT audi!” She literally turned me off to Oticon Intent with her lack of interest, commitment or even caring to spend time with the settings for me. I find it quite intriguing that so many others here do REALLY well with Oticon aids, and the Intent has helped them comprehend speech better than earlier models, too.

Since I get my hearing aids from the VA I have some experience with “less than perfect” audiologists. I try to stay informed about what is current in the hearing aid world so I can ask cogent questions, while not coming off as an “expert”.

@1Bluejay said:

My own anecdotal experience: I trialed the Intents for 2 weeks a few months back. Granted, I had an audi with the patience of a peanut setting them up. But I was actually annoyed at how LITTLE I understood speech right in my own house. We had some work done inside, and the tech guy would be 5-6’ away and I was simply UNABLE to even figure out what he was saying.

Strangely, I had a very similar experience with the Intents. Hearing in noise was amazing, but ordinary speech comprehension in ordinary situations such as the one you describe was surprisingly inconsistent. I’m thinking of going back for another trial with a different Audi.

@DaleM1 said:

But I’ve read that the spherical speech in loud noise doesn’t work well with a recorded track in which the speech and noise are combined.

Today I have found that video (from 5:55), so maybe it’s better than we supposed::

To answer the original posters question somewhere on this forum, it has been suggested to turn off “Bluetooth always connected” in the MyPhonak app due to the drain of the iPhone battery.

@Bimodal_user said:

Today I have found that video (from 5:55), so maybe it’s better than we supposed::

Thanks. I have also noticed (anecdotally) the same thing.

@DaleM1 said:

I expected my scores to improve over time as I unconsciously “learned” the word list, so the experiment design gives a slight bias to the Intents.

Using the same word list over again does seem like a test weakness; the best test would be double-blind with the words randomized between each test, if not a completely new set of words for each test, i.e., ideally, you should get someone else to make up the recordings and the written-out checklist of what the actual words were in the recordings. Still, a very interesting experiment, which roughly jives with what Phonak claims will be the improvement in understanding, IIRC.

@DaleM1 said:

So you’re really trading sound localization and occlusion reduction against music streaming and understanding speech in noise.

I deliberately wear ReSound Omnias with M&RIE receivers and occlusive fit pretty much for the reasons you state. The M&RIE receivers give me excellent localization (YMMV), and depending on one’s adjustability, one can habituate to an occlusive fit (it never bothers me).

Of course, with Spheric noise reduction, the Spheres can’t be beat so I’m looking forward to trialing them and breaking open my piggybank if I find them as good as everyone else says. I’d wear them with an occlusive fit, too.