[hts-users:01231] Re: HMGenS Synthesis with defined state duration

["Heiga ZEN (Byung Ha CHUN)" <zen@xxxxxxxxxxxxxxx>]

Hi,

Javier Latorre wrote (2008/03/17 13:43):

>  I wanted to ask whether in HMGenS is possible to define not only the 
> duration of the phones but also the duration of the states. From the 
> help output it seems that the "-s" options is exactly for this but I 
> could not find the way to define the format of the input MLF files. I've 
> tried something similar to the output duration file, that is
>
> #MLF!#
> "*/myfile.lab"
> 0 100 a=b-e+f#r.state[2]
> 100 200 a=b-e+f#r.state[3]
> 200 300 a=b-e+f#r.state[4]

The format is like as follows:

0 100000 sil[2] 113.748253 sil
100000 1600000 sil[3] 1626.428467
1600000 1650000 sil[4] 41.388893
1650000 1700000 sil[5] 48.251003
1700000 2500000 sil[6] 697.520630
2500000 2700000 ts-ch+i[2] 224.647842 sil-ch+i
2700000 3150000 ts-ch+i[3] 388.984680
3150000 3300000 ts-ch+i[4] 78.549385
3300000 3350000 ts-ch+i[5] 37.296726
3350000 3450000 ts-ch+i[6] 77.376114
3450000 3500000 ch-i+f[2] 47.528683 ch-i+s
3500000 4300000 ch-i+f[3] 1276.427246
4300000 4400000 ch-i+f[4] 130.119736
4400000 4550000 ch-i+f[5] 194.074692
4550000 4650000 ch-i+f[6] 116.448189
4650000 4750000 i-s+a[2] 44.935513 i-s+a
4750000 5050000 i-s+a[3] 230.516800
5050000 5350000 i-s+a[4] 308.950806
5350000 5500000 i-s+a[5] 103.292816
5500000 5600000 i-s+a[6] 77.362274
5600000 5750000 t-a+n[2] 233.337418 s-a+n
5750000 5850000 t-a+n[3] 183.698792
5850000 6000000 t-a+n[4] 284.620300
6000000 6050000 t-a+n[5] 87.169769
6050000 6100000 t-a+n[6] 70.149414

This is the format which HVite with "-a -f" option outputs.

Regards,

Heiga ZEN (Byung Ha CHUN)

--
------------------------------------------------
Heiga ZEN     (in Japanese pronunciation)
Byung Ha CHUN (in Korean pronunciation)

Department of Computer Science and Engineering
Nagoya Institute of Technology
Gokiso-cho, Showa-ku, Nagoya 466-8555 Japan

http://www.sp.nitech.ac.jp/~zen
------------------------------------------------