search

davircarvalho / Individualized_HRTF_Synthesis

Synthesis of individualized HRTFs based on Neural Networks, Principal Component Analysis and anthropometry
GNU General Public License v3.0
62 stars 7 forks source link

HRTF has amplified high frequrencies #12

Open ghost opened 11 months ago

ghost commented 11 months ago

Hello, I am really impressed by positioning in the HRTF that app makes. However, I have noticed that when I activate this HRTF, the high frequencies are significantly amplified, which can sometimes be unpleasant to listen to. Is it possible to obtain an HRTF with minimal distortion of the frequency response? I am currently using this HRTF through HeSuVi with AutoEQ.

ghost commented 11 months ago

I apologize in advance if my understanding of the usage or functioning of this program is not entirely accurate. It is possible that my question may seem foolish to some.

davircarvalho commented 11 months ago

hello @blwckcore some people have reported a similar experience, it could be that the dimensions you used are an edge case for the neural network. Would you mind to share the parameters you've used to generate the HRTFs so I can investigate further.

ghost commented 11 months ago

sadly, I've deleted screenshot with the data and the program with the runtime too. only things I have is the photo of my head and ear. I can remeasure them if you send the screenshot of your app with needed dimensions to measure (d1, d2, etc.).