search

r9y9 / nnmnkwii

Library to build speech synthesis systems designed for easy and fast prototyping.
https://r9y9.github.io/nnmnkwii/latest/
Other
393 stars 73 forks source link

Planning to attempt porting it to native C++ #117

Open SeleDreams opened 3 years ago

SeleDreams commented 3 years ago

Hi, just wanted to let you know that i'm thinking about working on porting the code of this and sinsy to C++ on my own in order to make it easier to integrate to projects If I end up getting it working i could send updates about it

because right now the dependencies with cython + numpy etc make it hard to port to other platforms than linux

SeleDreams commented 3 years ago

image I started working on it, however i'm kind of struggling to find a c++ equivalent of the interp1d function

edit: after checking i noticed only slinear is used in the lib anyway so I might in the meantime use a linear interpolation instead of a variable one

r9y9 commented 3 years ago

Cython and NumPy should work on windows without significant effort IMO. That being said, I understand that there are some demands on C++ implementation for embedded applications.

r9y9 commented 3 years ago

In typical DNN-based parametric TTS systems, linear interpolation is used to interpolate F0s for unvoiced regions. There's some work using spline interpolation but I think linear interpolation is good enough in my experience.

SeleDreams commented 3 years ago

Cython and NumPy should work on windows without significant effort IMO. That being said, I understand that there are some demands on C++ implementation for embedded applications.

My concern was more because of iOS for instance that has high restrictions on python and co, also the issue on windows is that scipy and all struggle to compile with mingw and require the proprietary visual studio compiler