Today I did my yearly test of my smoke detectors. With my Android-App I can see yelling (90dB@3feet) sounds at 3.3 to 3.5 kHz depending on the instrument being tested. Without HA I don’t hear any sound at all. With my HAs turned on the sound is moved to 550-750 Hz. At least it seems to be compared to playing the note what I hear on a piano. Hm interesting because Phonak translates high frequencies down to at lessest 800Hz. Looks like there are two interferencing frequencies resulting in this low frequency.
I want to find a very small, but accurate microphone and couple it to one of my receivers so I can see exactly how SR2 does its thing. I have access to equipment that would let me plot the input frequency vs output frequency. I could adjust the settings of SR2 and then see how they affect the output. I found it interesting that although I can hear as high as 7.5 kHz through SR2, it doesn’t really help with speech all that much. All of the major components of speech are supposed to be contained within 300 Hz to 3.3 KHz because that was the frequency response of telephones back in the analog days.
I would like to hear how that experiment goes. Would you have a recording or some noise that can be consistently played to use for comparison?
It has taken me wearing the aids every day, all day to see the speech recognition improvement. It’s still not great but much better than it was. Just now the coffee maker made some beeps to let us know it had finished making coffee. So many new sounds I have never heard.
1 Like
3.3khz will get you most speech sounds but won’t get you fricatives like s,f and th, especially with female speech. The “target” for frequency lowering with the goal of speech recognition is to make about 6.3khz audible. Frequency lowering is definitely a compromise. I seemed to tolerate it best if I made gradual changes and allowed myself to get used to it.
1 Like
I found a mic: Phonak’s original lapel mic for compilot 
Attached you find my first try to record a linear sweep beginning at 100Hz and ending at 8kHz within 4 seconds. The upper figure is the recording with the HA turned off. The lower is my Phonak Naida B. Well, the mic gets too much input from source instead of from the HA itself, but I find it interesting anyway. I have to find a way to heavily reduce leaking volume from source.
1 Like
I’d like to know your interpretation of that. A lot of audiology equipment is online (maybe because a lot of audiologists quit) so you can find things very cheaply sometimes. Some of the gear comes with a soundproof box you put your aid in and you can even couple the aid physically using a silly-putty type molding clay (no idea what it’s really called). I bought an [audioscan verifit for a few hundred bucks off eBay.
What was Sound Recover set at in this example? SR2?
It looks like you could dismiss everything in the upper frequencies. Just guessing but appears to peak at 3000 hertz.
That’s interesting. I can actually hear the rain hitting our metal roof now. Couldn’t before.
1 Like
Thank you for the info but that’s way too much for me just for fun.
My interpretation of my figure: I did not find good settings to convert the wavefrom to a spectrogram. The only interesting information is below the original source: That’s where the HA transforms the input waveform to. There you see many steps after each few hundret Hz but I don’t believe that this is real. It highly depends on the settings to transform to a spectrogram. The previous figure uses a frequency transformation using Hanning with window-size 4096. The figure attachted next uses Gaussion(a=4.5) with window-size 16384. I have scaled the output to the range what I can hear: 0-2kHz. And indeed there is a gap around 2.6kHz where I don’t hear anything with HA. I had expected it to be 2.0-3.0kHz (@Raudrive I am using SoundRecover 2 set to “3d”, i.e. starting at 1.1kHz, cut-off2/maxout=3.0kHz). You can see that the starting point is correct. And you can see that I can hear some resonant sounds at 0.6kHz like my smoke detector test showed up this morning. Anyway, this is not a science - I was just interested after John’s post.
2 Likes