search

bishoph / sopare

Real time sound pattern recognition in Python for Raspberry/Banana Pi.
Other
321 stars 86 forks source link

Wrong spoken word. #83

Open kimgenegaby opened 4 years ago

kimgenegaby commented 4 years ago

I noticed you concentrated lots of time in YouTube in the frequency domain. How to get frequency domain with a wav file? Never mind. I compiled the Fscape Java program from Github( by three way, this contributor is also from German. Right?) to have frequency domain. Another question I have is that "sorted_best_match" was 98%, but the "match_result" shows {u'test'} not the right spoken word as you had in the YouTube.

bishoph commented 4 years ago

Not sure what happened but maybe the words are very similar and the best prediction was used and displayed...

kimgenegaby commented 4 years ago

Hello Bishoph,

 Thanks Bishoph for the fast response. I am very grateful for you to give out the free Github python programs. Is there a way to give a donation?
 After you showed the video referring to frequency transform of the microphone, I might have a problem with my microphone after I frequency transformed my recorded message. Therefore, I will buy a microphone that has better recording quality knowing I don't have to rely on online Google's speech to text.

Again thank you.

Steve

bishoph commented 4 years ago

Glad you find my work useful. There is currently no way for donations - but I'll think about it :)

Have fun!

kimgenegaby commented 4 years ago

Hello Bishoph,

 When you are ready for a donation,

just email me!

Have a good day!

Steve