Dicky
Dicky
MIDI Play and keyboard test base on Timidity and ImGui GitHub project: Tanlu libmidi https://github.com/tanluteam/libmidi.git
@aloneguid, first I need calculate audio data to FFT data and compose to db value, then I create a buffer to maintain every buffered window(in my case, it is 1024...
try to using full path for istats such like /usr/local/bin/istats
meet same problem, seems MatMul layer cause segmentation. I think is it ncnn's tools bug.
could you share those kind of strategies? It is very interesting result for me.
yes, new result is much better then simple cos distance. I am eager for new method sharing.
I meet a problem when I convert vit-h encoder model: Traceback (most recent call last): File "scripts/export_image_encoder.py", line 183, in run_export( File "scripts/export_image_encoder.py", line 167, in run_export weights = np.frombuffer(fp.read(),...
@csukuangfj Thanks!!! Because audio algorithm using frequency domain data generally, so I asked. wave-reader is simply get waveform data from file, and then send data into knf::OnlineFbank. What my real...
I understood now. You are using fbank as audio feature extract, including fft. low pass filter, mcl filter. Then send filterd data into ncnn. Thanks for you anwser, it is...
we using full screen window for application. and at left side menu toolbar. Last button is quit button, you can click it to exit app.