Hi,
I am new in the domain of speech synthesis (at least in practice), and I would like to build an hts voice from the data of my lab. I first built a voice with the HTS-demo-CMU-Arctic in order to understand how it works, but now I try to use the scripts of this demo to build my own voice. I have .lab and .raw files, and I changed the makefiles and script of the demo, but I cannot build the voice. During the first steps (MGC coefficients extraction, lf0 extraction ...), no file is created, so I cannot go further.
These data have already been used to build a voice in the past, and I have some other files which have certainly been created but I don't know how, I don't know what they are and I don't know how to use them. Here is the list of these files :
beep_def.ch,
f2b_features_maps.ch,
f2b_lrdur.ch,
f2b_tobi_lrf0_model.ch,
files, mjb.about,
mjb.chatr98,
mjb.concat,
mjb_discretetables.ch,
mjb.lbg,
mjb_lrdurstats.ch,
mjb.phonetable,
mjb.prosody,
mjb.sdf,
mjb.tdf,
mjb_tobif0.ch,
mjb.udb,
mjb.vqdist,
mjb.wavlib,
mjb_weights0.ch,
pause_prediction_tree.ch,
tobi_accent_tree.ch,
tobi_boundary_tone_tree.ch
Could anyone tell me how I could use these files, or how to build the voice using raw and labels only ? Thanks
Pierre-Edouard