Difference between revisions of "Text To Speech"

Latest revision as of 08:15, 3 May 2010

{Stub needs to be expanded on how to add more voices to get around licensing issues}

As of LMCE 0810, Text To Speech will now use the festival speech engine. This will allow the use of more human sounding voices in the future. Also as of LMCE 0810, you can now specify the voice to use for your announcements.

Here is an example of how to use Text To Speech: 1) Create a scenario. In this example, I'm just going to create a climate scenario called "Test" 2) From the edit scenario page of your new "Test" scenario, use the dropdown to edit using the Advanced wizard 3) Add a new command by using the Device drop-down to select "Text To Speech" 4) You will now get a box where you can select a command to use. Select the "Send Audio To Device" command. 5) You will now get a list of command parameters to fill out:

1. 103 List PK Device - enter a comma delimited list of devices to send the audio to. For our test, just look up the device number of your On Screen Orbiter, and enter it here.
2. 9 Text - Enter your text here, such as 'Hello. This is a test. A B C 1 2 3'. Notice the single-quotes - they must be used!
3. 278 Voice - Here is where you can select any one of the installed voices. The default installed voices are:

voice_lp_diphone
voice_pc_diphone
voice_el_diphone
voice_suo_fi_lj_diphone
voice_hy_fi_mv_diphone
voice_marathi_NSK_diphone
voice_don_diphone
voice_ked_diphone
voice_rab_diphone
voice_kal_diphone (this is the default voice)
voice_telugu_NSK_diphone
voice_czech_ph
voice_hindi_NSK_diphone

just type in the voice you want to use. If you enter the voice incorrectly, you will get no sound.

Setting a default voice

Browse to the Text_To_Speech device by switching to the devices tree view on the left pane in the web admin. As of 0810, you can now specify a default voice to be used with the "default voice" device data. Enter the desired voice in the same format listed above for the Send Audio To Device command.

Some better sounding voices for testing

These voices are made by the HTS working group hosted at the Nagoya Institute of Technology. Some of these are very good for free voices! 1) Download the voices:

cd ~/
mkdir hts_tmp
cd hts_tmp/
wget -c http://hts.sp.nitech.ac.jp/archives/2.1/festvox_nitech_us_awb_arctic_hts-2.1.tar.bz2
wget -c http://hts.sp.nitech.ac.jp/archives/2.1/festvox_nitech_us_bdl_arctic_hts-2.1.tar.bz2
wget -c http://hts.sp.nitech.ac.jp/archives/2.1/festvox_nitech_us_clb_arctic_hts-2.1.tar.bz2
wget -c http://hts.sp.nitech.ac.jp/archives/2.1/festvox_nitech_us_rms_arctic_hts-2.1.tar.bz2
wget -c http://hts.sp.nitech.ac.jp/archives/2.1/festvox_nitech_us_slt_arctic_hts-2.1.tar.bz2
wget -c http://hts.sp.nitech.ac.jp/archives/2.1/festvox_nitech_us_jmk_arctic_hts-2.1.tar.bz2
wget -c http://hts.sp.nitech.ac.jp/archives/1.1.1/cmu_us_kal_com_hts.tar.gz
wget -c http://hts.sp.nitech.ac.jp/archives/1.1.1/cstr_us_ked_timit_hts.tar.gz

2) Unpack the voices:

for t in `ls` ; do tar xvf $t ; done

3) Install the voices:

sudo mkdir -p /usr/share/festival/voices/us
sudo mv lib/voices/us/* /usr/share/festival/voices/us/
sudo mv lib/hts.scm /usr/share/festival/hts.scm

4) Clean up

cd ~/
rm -rf hts_tmp/

you can now go to a prompt, and enter festival interactive mode:

festival

then get the new list of voices:

(voice.list)

you can use any of these voices by prepending with "voice_". My favoirite and the most human sounding is: voice_nitech_us_slt_arctic_hts