Festival 2.4 Regression

[WhyteHorse]

Festival 2.4 breaks compatibility with HTS voices. Apparently the developers break compatibility every time they release a new version and it no longer works with the older version's HTS voices, but for a fee, they will let you use their proprietary voices... wink wink nod nod.
To reproduce, follow the very old instructions here to add an HTS voice:
https://ubuntuforums.org/showthread.php?t=751169&page=18

Then run festival at the command line and enter this:

festival> (voice_nitech_us_slt_arctic_hts )
nitech_us_slt_arctic_hts
festival> (SayText "Hello from Ubuntu")
Warning: HTS_fopen: Cannot open hts/htsvoice.
aplay: main:593: bad speed value 0
#<Utterance 0x7f3b2f4aaa90>

I think we should roll back to a known-good-working festival(2.1?) with HTS voices, fork that version, patch it with the bug fixes in 2.4 less the code that breaks it, and move on. There's a patch available here:
https://aur.archlinux.org/packages/festival-patched-hts/

ProblemType: Bug
DistroRelease: Ubuntu 16.04
Package: festival 1:2.4~release-2 [modified: usr/share/festival/hts.scm usr/share/festival/voices.scm]
ProcVersionSignature: Ubuntu 4.4.0-45.66-generic 4.4.21
Uname: Linux 4.4.0-45-generic x86_64
ApportVersion: 2.20.1-0ubuntu2.1
Architecture: amd64
CurrentDesktop: Unity
Date: Fri Oct 28 07:48:58 2016
InstallationDate: Installed on 2014-09-03 (785 days ago)
InstallationMedia: Ubuntu 14.04.1 LTS "Trusty Tahr" - Release amd64 (20140722.2)
SourcePackage: festival
UpgradeStatus: Upgraded to xenial on 2016-09-29 (29 days ago)
mtime.conffile..etc.festival.scm: 2016-10-17T19:31:25.018208

[WhyteHorse (whytehorse)]

Workaround: Install festival from source, download voices from festival and install them, and voila! It works.

[WhyteHorse (whytehorse)]

FYI here are the install from source instructions:
https://github.com/usc-isi-i2/festival-text-to-speech-service/blob/master/README.md

[WhyteHorse (whytehorse)]

After installing festival, you need to make a symbolic link to it. On my system the command is:
sudo ln -s /home/ben/festival/festival/bin/festival /usr/bin/festival

[Sergio Oller (zeehio)]

Hi,

Feel free to use the latest Festival-2.4 from Ubuntu repositories with the festvox-us-slt-hts package currently at debian/sid that packages the voice you mention and at package build converts it to festival-2.4 format.

https://packages.debian.org/sid/festvox-us-slt-hts

[Sergio Oller (zeehio)]

Ubuntu zesty has already synced the festvox-us-slt-hts package http://packages.ubuntu.com/zesty/festvox-us-slt-hts

[whitis (whitis)]

This is marked as fixed and yet, on ubuntu 18.04.3 LTS, installing the festvox-us-slt-hts does not install any of the NITECH voices (only a single CMU voice) and if you try to install them using the directions in the nitech section at https://ubuntuforums.org/showthread.php?t=751169, festival fails with the same error when you try to use the nitech voices.

festival --version
festival: Festival Speech Synthesis System: 2.5.0:release December 2017

[Sergio Oller (zeehio)]

As far as I can remember the HTS voice in `festvox-us-slt-hts` is the same from nitech.

If you want to convert any nitech voice from the old format to the new, you can do it in the same way that it is done within the festvox-us-slt-hts package. So far no one has bothered to do it, probably because no one has the time, but it should work.

To do the conversion, you should first follow the debian/rules file, that uses a python script to create the .htsvoice file, and second applying similar patches to the ones available in debian/patches, to fix the festvox/* files accordingly.

I think it is a much better solution to migrate the old voices to the new system, if possible, than to patch the system and maintain a fork of an already old and not very well maintained piece of software, but do as you please.