Hi,
I did that a while ago as part of a contribution to the Debian TTS team.
Under Debian, Ubuntu and derivatives you can use the festvox-us-slt-hts available in debian/sid [1] (just download the package from the website and install it if you want).
I actually used a python script [2] (written by Peter Drysdale, another Deabian TTS team member) to convert the old hts voice to the new format, but you won't need that.
As you already have the new htsvoice file, you need to:
1. Download the festival 2.1 hts voice from [3]
2. In the festival/lib/voices/us/cmu_us_slt_arctic_hts/hts directory remove all files except feat.list
3. Copy your htsvoice file into the directory festival/lib/voices/us/cmu_us_slt_arctic_hts/hts (so you will end up having two files: cmu_us_slt_arctic_hts.htsvoice and feat.list)
4. Edit festival/lib/voices/us/cmu_us_slt_arctic_hts/festvox/cmu_us_slt_arctic_hts.scm. You need to set the following hts_engine_params:
(set! cmu_us_slt_arctic_hts::hts_engine_params
(list
(list "-m" (path-append hts_data_dir "cmu_us_slt_arctic_hts.htsvoice"))
'("-g" 0.0)
'("-b" 0.0)
'("-u" 0.5)
))
5. Check that the .htsvoice file name matches the one in festvox/cmu_us_slt_arctic_hts.scm
If you have doubts check how the debian package does it, it's working well there.
In case you want to try, there is also `mimic` [4], a flite fork for the Mycroft AI [5] that will have support for hts voices by the end of this week (or next week) [6].
Best,
Sergio Oller