Phonak Infinio Sphere vs Oticon Intent – quantitative word recognition testing

Hearing Aids
,
#21

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:

sample audio files for speech recognition (kaggle.com)

After denoising by Adobe Podcast AI:
https://file.io/h7WaVhTn2moC

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

image
image1920×1429 160 KB

#22

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.

2 Likes
#23

Nice experiment Dale, thanks for sharing.

#24

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).
1 Like
#25

Wow, quite a scientific approach! Thanks for testing.

I other words, regarding SNR and intelligibility:

lowest → medium → highest

original → Spheric prog. → Adobe Podcast AI.

Yes, I had to upload the file, and it required a few moments to proceed with AI denoising. However, it seems that the Spheric Speech Clarity program works better than expected with a recorded track in which the speech and noise are combined.

#26

There is a note to HearAdvisor’s independent comparison of Phonak Infinio Sphere and other premium hearing aids including Oticon Intent. your finding is quite in line with their finding, amazing.

#27

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

#28

Wow, that was creative…

#29

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.

#30

Your loss is reasonably flat from 250 Hz to 8000 Hz - you should try the AirPod Pro 2 when apple releases the self program. Retest with them in the ear and that may tell you how to tweak the programming

#31

Yes thanks may well be worth a try.

#32

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. :slight_smile:

1 Like
#33

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”.

#34

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.

1 Like
#35

This is great feedback, and great to see people taking initiative to truly put these devices to the test.

#36

@DaleM1, it you like scientific approach and experimenting, there is probably ideal page and app:

https://wwwn.cdc.gov/niosh-mining/HLSimWeb

If you feel like it, you can test denoising by Sphere AI here.

1 Like
#37

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

#38

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.

#39

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

1 Like
#40

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.

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.

1 Like