I'm running Debian Linux (specifically, an up-to-date version of the
"testing" distribution). I installed festival via dpkg and downloaded
cmu_us_slt_arctic_hts.tar.gz and cmu_us_awb_arctic_hts.tar.gz.
I unzipped these files, and moved the "us" directory to my
/usr/share/festival/voices directory (which already existed and had other
voices installed in directories named "english" and "spanish")
However, when i try to test, i get the following:
festival> (voice_cmu_us_awb_arctic_hts )
cmu_us_awb_arctic_hts
festival> (SayText "Test.")
Cannot open file /tmp/est_30819_00000/utt.wav as tokenstream
Wave load: can't open file "/tmp/est_30819_00000/utt.wav"
Cannot load wavefile: /tmp/est_30819_00000/utt.wav
#<Utterance 0x4082a328>
I even recompiled hts_engine and copied it to
/usr/share/festival/voices/us/cmu_us_slt_arctic_hts/hts and
/usr/share/festival/voices/us/cmu_us_awb_arctic_hts/hts but i still get the
same problem. I made sure that the binary looked okay:
$ /usr/share/fl/voices/us/cmu_us_awb_arctic_hts/hts/hts_engine
hts_engine - A compact HMM-based speech synthesis engine
usage:
hts_engine [ options ] [ infile ] > stdout
[...]
Did i follow the right installation procedure? Has anyone else seen this
problem? Any suggestions for figuring out the root cause of the problem?