Alok Prasad

@MlWoo Can we add this in Tacotron training for silence removal https://github.com/gooofy/zamia-tts/commit/66bd10d4c34f215eaf37fba7e712047291002ace

@superhg2012 i guess T1 and T2 are same except Vocoder part , which anyways we are using LPCNET. Can you share the changes for T1 with LPCNET.

nb_features is already 55 , so u mean to say no changes in lpcnet just train lpcnet. https://github.com/mozilla/LPCNet/blob/master/src/train_lpcnet.py For T1 Instead of num_mels = 20 you mean num_mels = 55?

> @byuns9334 I don't get good quality with T2 + LPCNet(20dim). But, I get better quality with T1 and LPCNet(55 dim). > > @lmingde I put the dumped f32 files...

how is test.lab created?

looks like Merlin was used to generated the lab files *.lab: label files (time-to-phone alignments) http://jrmeyer.github.io/tts/2017/02/14/Installing-Merlin.html

You can try c++ project https://github.com/alokprasad/lpctron-tts-cpp it has model in model folder ( you to merge the files to generate the pb). Will upload samples later.

can you modify Cmakelists.txt uncomment include(CMakeLists-static-linux.txt) and comment the android one.

@terryyizhong what is linux and machine it is ubuntu / X86?

@ysujiang i will upload sample soon , but did u trained the lpcnet also ?