-
Till now for consistency all our *Bands algorithms (MelBands, ERBBands, BarkBands, FrequencyBands) output energies while MFCC and GFCC output log-energies. In addition TriangularBands has an option to…
-
Librosa gives us MFCCs - we should use those.
tkell updated
7 years ago
-
A) I managed to clean up the code and replicate all the results but one for the 2013 dataset (supervised : late/early ; mfcc/scattering/log-scattering).
I made a mistake when selecting the optimal …
-
Is it possible to run `pocketsphinx_continuous ` command on GPU? Is there any utils or prepared layers for that? Thanks
-
Hello,
would be nice if you could extend the MFCC example script (https://github.com/aubio/aubio/blob/master/python/demos/demo_mfcc.py) to also calculate delta and double-delta coefficients.
Thanks!…
-
Does eesen convient to extract MFCC and fbank with first order difference?
-
Hi all !
I'm new here, I'm trying to train a RNN using LSTM for phoneme classification. The input to the RNN is the 39 MFCCs coefficients of the current time frame, and the output is one of 39 phoneme…
-
add
- parameter to select DCT type III (used by htk) or DCT type II (use in Slaney's). Currently in essentia we have type II . Note that in librosa they use type III by default.
- paramter for op…
-
Hello, What method would you suggest for training the MFCC and how to setup feature vectors?
Should I implement a state by state approach for each sample?
-
We need an example scripts somewhere for recent nnet recipes (nnet2 and nnet3) with pitch features, that are suitable for online decoding. E.g. in hkust or gale_mandarin. I may have to do this mysel…