Implementing Band Energy Ratio in Python from Scratch

[@kitulinokitulino3360]

Excellent, thank you Valerio grazie mille for the amazing series ! One comment if you allow me. With numpy, here is a better way to calculate sum() on matrix instead of using normal python lists for-loop. band_energy_ratio = power_spectrogram[:split_frequency_bin,:].sum(axis = 0) / power_spectrogram[split_frequency_bin:,:].sum(axis = 0)

[@ValerioVelardoTheSoundofAI]

Glad you liked the series! Thank you for the suggestion :)

[@louisleeboy]

Very useful video, save me a lot of time for looking up the information and practice it. Thanks.

[@facefamousee]

nice video. I've learnt a lot! Thank you.!

[@evafrass9465]

Very good. Thank you so much. 👌

[@amirhosseinsorour836]

Thank you for your useful videos. Is there any built-in function that calculates BER so that we don't need to implement it from scratch?😁 Like the ones in librosa for calculating ZCR, RMS, STFT, etc.

[@jeremynx]

Thank you for such informative and clear videos!

[@ValerioVelardoTheSoundofAI]

Thank you!

[@haowu3319]

should the frequency_delta_per_bin = frequency_range / (spectrogram.shapre[0] - 1) ? Since the spectrogram.shapre[0] is the number of the bins from freq zero to nyquist freq. The freq segments between freq zero to nyquist freq is (spectrogram.shapre[0] - 1)

[@mamdouhdabjan9292]

Hey man, I have noticed an error in the GitHub code. Unfortunately, I cannot propose the changes due to restrictions but in the GitHub code the function that locates the bin that has the threshold frequency uses the number of frames as a locater instead of the number of bins in a frame. You can use spectrogram.shape[0] instead of len(spectrogram[0]). Thanks for the amazing content

[@markusbuchholz3518]

Hello Valerio, thanks for your video. Very impressive as always. The whole channel and especially this series is very impressive. From my point of view you can consider (probably it very time consuming) to summary your "lectures" in the interactive book (sorry but I can not remember it can be that you have started and you will publish your great book soon). Anyway if you have not started it can be useful to use www.kickstarter.com in order to receive financial support for your great "project". I am sure there will be a good response from community and other institutions. Of course, you can rely on me - heeee. Have a nice day!

[@ValerioVelardoTheSoundofAI]

Thank you Markus :) As you might know, I've planned to work on a book (DL + Audio), but, at the moment, I'm struggling a bit with time. I've decided to release "short" videos with self-contained cool AI Audio applications (with a format similar to the "Make a Spotify Playlist in Python"). These videos will provide the core for the chapters of the book, which nonetheless I'll have to write up! I'm thinking of opening a Patreon account when the community grows a bit more. Kickstarter is definitely an option I'll also consider in the future.

[@tariqahassan5692]

Nice lesson, but we barely can see the text.. very tiny text font