[Subject Prev][Subject Next][Thread Prev][Thread Next][Date Index][Thread Index]

[hts-users:03013] Re: Checking synthesized speech in demos


Hi Dechen,

Thank you for the quick reply.

I checked about my sox version on the sox mailing list and I seem to have all the relevant support except for alsa which seems to be a native linux format, and I am working on Cygwin in Windows. I have also used my version of sox to convert wav files to raw files, so it does not look like the problem is with sox. If the alsa support is required then I need to switch to Linux.

I added a print command to training.pl as specified in http://hts.sp.nitech.ac.jp/hts-users/spool/2011/msg00001.html to try to figure out what was happening. My log now gives the following error:

Error: HTS_get_fp: Cannot open /cygdrive/c/Huda/TTS/Build/HTS-demo_CMU-ARCTIC-SLT/voices/qst001/ver1/dur.pdf.

So it seems that my dur.pdf files are not getting prepared, but I don't seem to be getting any errors before this. I could not find any information on the archives about this, so if anyone can give me hints on how to investigate this further it would be much appreciated.

--
Thanks & regards,
Huda Sarfraz
www.cle.org.pk


On Thu, Aug 18, 2011 at 1:28 PM, dechen chhoeden <dnamgyelster@xxxxxxxxx> wrote:
Hi,

Looks like "sox" is not installed properly. Sox is supposed to convert various formats of computer audio files into other formats (in our case to .wav files). In your log file, the error posted "C:\cygwin\bin\sox.exe FAIL formats: can't open output file `/cygdrive/c/Huda/TTS/Build/HTS-demo_CMU-ARCTIC-SLT/gen/qst001/ver1/1mix/0/alice01.wav': No such file or directory" potentially points to that. Why don't you reinstall 'sox' and re-train HTS. [sox must have these package/library files after installation: libsox_fmt_alsa.so  libsox_fmt_gsm.so    libsox_fmt_vorbis.so
libsox_fmt_flac.so  libsox_fmt_lpc10.so  libsox_fmt_wavpack.so
]
Good luck.

Regards,
Dechen


On Thu, Aug 18, 2011 at 8:56 AM, Huda Sarfraz <huda.sarfraz@xxxxxxxxxxx> wrote:
Hi,

Thanh-Son PHAN, I have checked my log file and there seems to be an error in waveform synthesis.

Dechen, as far as I can tell, required files have been generated during the training. My project/gen/qst001/ver1/1mix/0/ (0,1 and 2) folders have dur, lf0, mgc, pit and unv files for alice01 through alice40 after training.

I've looked up the archives and cannot find any clue, so any hints would be appreciated.

Here is the error at the end of the log file, it says that it cannot open the wav file, but isn't it supposed to be creating the wav file?

=====================================================================
Start synthesizing waveforms (1mix) at Thu Aug 18 06:50:01 TST 2011
=====================================================================

Processing directory /cygdrive/c/Huda/TTS/Build/HTS-demo_CMU-ARCTIC-SLT/gen/qst001/ver1/1mix/0:
 Synthesizing a speech waveform from alice01.mgc and alice01.lf0...C:\cygwin\bin\sox.exe FAIL formats: can't open output file `/cygdrive/c/Huda/TTS/Build/HTS-demo_CMU-ARCTIC-SLT/gen/qst001/ver1/1mix/0/alice01.wav': No such file or directory
Error in /usr/bin/excite -p 240 /cygdrive/c/Huda/TTS/Build/HTS-demo_CMU-ARCTIC-SLT/gen/qst001/ver1/1mix/0/alice01.pit | /usr/bin/dfs -b -0.0055231 -0.00983208 -0.0087294 -0.000428110000000001 0.01195534 0.0207699 0.0177785 0 -0.02604947 -0.04502049 -0.03915007 0.00219802999999999 0.07492151 0.15919616 0.22659094 0.25231954 0.22659094 0.15919616 0.07492151 0.00219802999999999 -0.03915007 -0.04502049 -0.02604947 0 0.0177785 0.0207699 0.01195534 -0.000428110000000001 -0.0087294 -0.00983208 -0.0055231 | /usr/bin/vopr -a /cygdrive/c/Huda/TTS/Build/HTS-demo_CMU-ARCTIC-SLT/gen/qst001/ver1/1mix/0/alice01.unv | /usr/bin/mglsadf -m 34 -p 240 -a 0.55 -c 0 /cygdrive/c/Huda/TTS/Build/HTS-demo_CMU-ARCTIC-SLT/gen/qst001/ver1/1mix/0/alice01.mgc | /usr/bin/x2x +fs -o | /usr/bin/sox -c 1 -s -2 -t raw -r 48000 - -c 1 -s -2 -t wav -r 48000 /cygdrive/c/Huda/TTS/Build/HTS-demo_CMU-ARCTIC-SLT/gen/qst001/ver1/1mix/0/alice01.wav



On Wed, Aug 17, 2011 at 6:32 PM, dechen chhoeden <dnamgyelster@xxxxxxxxx> wrote:
Hi Huda,

Ya the generated speech files should be in project folder/gen. Since you are trying to run just the original HTS demo (HTS-demo_CMU-ARCTIC-SLT), it has all the necessary training data. Hence you just have to run it. May be your "data" was not built properly. Why don't you check the "Makefile" of project folder/data and see if all the required folders and files were generated while training data (like lf0, cmp, scp, mcp, cmp, lists and scripts => these folders and their respective files will be generated while training 'data'). In the main Makefile in the project folder, uncomment voice and train just data first. Then only you can built voice. Just to make sure. Otherwise if these things where fine, then may be you are missing some installation files - make sure SPTK, ActiveTCL and other accompanying sws needed are installed properly. Hope this is of some help to you.

Regards,
Dechen


On Wed, Aug 17, 2011 at 12:18 PM, Huda Sarfraz <huda.sarfraz@xxxxxxxxxxx> wrote:
Hi,

I have run HTS-demo_CMU-ARCTIC-SLT and am trying to find the generated files to confirm that I have run it correctly. I found out that .lab files in data\gen are used for testing (http://hts.sp.nitech.ac.jp/hts-users/spool/2011/msg00220.html), but I cannot find the generated files in my project folder. I have looked in voices and gen, but there are no sound files.

Are the sound files placed somewhere else? Do I need to do something in addition to the install instructions to get synthesized speech from the .lab files in data\gen?

Any suggestions will be much appreciated.



Follow-Ups
[hts-users:03014] Re: Checking synthesized speech in demos, Huda Sarfraz
References
[hts-users:03007] Checking synthesized speech in demos, Huda Sarfraz
[hts-users:03008] Re: Checking synthesized speech in demos, dechen chhoeden
[hts-users:03010] Re: Checking synthesized speech in demos, Huda Sarfraz
[hts-users:03011] Re: Checking synthesized speech in demos, dechen chhoeden