# # 1 # # # # 11 # # ## # 1 1 ####### # # # 1 # # # # # 1 # # # ## 1 # # # # 11111 release 4 - 17 September 1999 -------------------------------------------------------------- HN1 - Hanmal (Korean language) male voice for the MBROLA synthesizer Created by : Prof. KIM, Kyongsok/GIM, Gyeongseog -------------------------------------------------------------- Table of Contents -------------------------------------------------------------- 1.0 Description of the HN1 diphone database 2.0 Installation and tests 3.0 Contacts -------------------------------------------------------------- 1.0 Description of the HN1 diphone database -------------------------------------------------------------- This database consist of 1909 diphones. The name for Hanmal diphone database is HN. Since, both South and North Koreas use the same language, it does not seem proper to use as a database name KR, which stands for South Korea only. Therefore, we will use HN, abbreviated from Hanmal (Korean Language). 'Han' stands for Korean and 'Mal' for Language. Note: According to ISO 3166-1, KR and KOR are for South Korea (Republic of Korea) and KP and PRK are for North Korea (Democratic People's Republic of Korea). For the same reason as in 4), we will use "Hanbando" (Korean Penninsula) flag for HN diphone database. The flag was first used at some international table tennis game in 1991 to represent a South-and-North-Koreas-unified team. The flag is also being used at ships which carries South Korean tourists going to Geumgangsan in North Korea. The flag shows the whole Korean Penninsula, i.e., both South and North Koreas. Software used for recording, labeling, etc: speech filing system (sfs) developed at Department of Phonetics and Linguistics, University College London, London, U.K: http://www.phon.ucl.ac.uk/resource/sfs.html SAMPA Description of Hanmal Romanization Symbols (1999.07.29, 10:00 KST) ------------------------------------------------ XSAMPA (Extended SAMPA) has been use to encode the database. Thanks to the mbrola renaming mechanism, one can use initialization files (or Mbroli renaming/mapping tables) to use an alternative phoneme set issued from romanization of Hangeul - MOE 1959 (Korean Ministry of Eduction, 1959 edition), which is nearly the same as ISO TR 11941: Transliteration of Korean script (Method II in case of Consonantal lettres.) Consonants (Onset Position) --------------------------- Roman SAMPA Description 101. g k velar lax plosive, voiceless 102. gg k_> non-pulmonic velar ejective 103. n n alveolar nasal 104. d t alveolar lax plosive, voiceless 105. dd t_> non-pulmonic alveolar ejective 106. r 4 alveolar tap 107. m m bilabial nasal 108. b p bilabial lax plosive, voiceless 109. bb p_> non-pulmonic bilabial ejective 110. s s alveolar fricative, voicelss 111. ss s_> non-pulmonic alveolar fricative ejective 112. ngo ?? ??? 113. j ts\ alveolo-palatal lax affricate, voiceless 114. jj ts\_> non-pulmonic postalveolar affricate ejective 115. ch ts\_h postalveolar aspirated affricate, voiceless 116. k k_h velar aspirated plosive, voiceless 117. t t_h alveolar aspirated plosive, voiceless 118. p p_h bilabial aspirated plosive, voiceless 119. h h glottal fricative, voiceless 120. gv g velar plosive, voiced 121. dv d alveolar plosive, voiced 122. bv b bilabial plosive, voiced 123. jv dz\ postalveolar affricate, voiced 124. sh s\ alveolo-palatal fricative, voiceless 125. lo l alveolar lateral approximant Consonants (Coda Position) -------------------------- Roman SAMPA Description 301. gc k_} velar plosive, voiceless 304. nc n_} alveolar nasal, no audible release 307. dc t_} alveolar plosive, voiceless 308. l l_} alveolar lateral approximant, no audible release 316. mc m_} bilabial nasal, no audible release 317. bc p_} bilabial plosive, voiceless 321. ng N velar nasal Vowels ------ Roman SAMPA Description 201. a a open front unrounded, Cardinal 4 202. ae E open-mid front unrounded, Cardinal 3 203. ya ja palatal approximant + open front unrounded 204. yae jE palatal approximant + open-mid front unrounded 205. eo V open-mid back unrounded 206. e e close-mid front unrounded, Cardinal 2 207. yeo jV palatal approximant + open-mid back unrounded 208. ye je palatal approximant + close-mid front unrounded 209. o o close-mid back rounded, Cardinal 7 210. wa wa voiced labial-velar approximant + open front unrounded 211. wae wE voiced labial-velar approximant + open-mid front unrounded 212. oe 2 close-mid front rounded 213. yo jo palatal approximant + close-mid back rounded 214. u u close back rounded, Cardinal 8 215. weo wV voiced labial-velar approximant + open-mid back unrounded 216. we we voiced labial-velar approximant + close-mid front unrounded 217. wi wi voiced labial-velar approximant + close front unrounded 218. yu ju palatal approximant + close back rounded 219. eu M close back unrounded 220. eui M\i velar approximant + close front unrounded 221. i i close front unrounded, Cardinal 1 -------------------------------------------------------------- 2.0 Installation and Tests -------------------------------------------------------------- If you have not copied the MBROLA software yet, please consult the MBROLA project homepage and get it: http://tcts.fpms.ac.be/synthesis Copy hn1.zip into the mbrola directory and unzip it : unzip hn1.zip (or pkunzip on PC/DOS) On Unix, try: ------------ mbrola hn1/hn1 hn1/TEST/s131.pho test.wav If you want to use the romanize phoneme set: mbrola -I hn1/TEST/hn1roman hn1/hn1 hn1/TEST/s131roman.pho test.wav It creates a short sound file, RIFF Wave format in this example. But depending on the extension test.au, test.aif, or test.raw other file formats can be obtained. Listen to it with your favorite sound editor, and try the other command files (*.pho) to have a better idea of the quality of speech that can be synthesized with MBROLA and the HN1 database. On Unix systems you can pipe the audio ouput to the sound player as on a HP : mbrola hn1 TEST/s132.pho - | splayer -srate 16000 -l16 Also refer to the readme.txt file provided with the mbrola software for using it. On Windows: ---------- Once hn1 has been registered in Mbroli, simply double click on TEST/s131.pho If you want to use renamed phoneme set, you must initialize Mbroli's diphone renaming/mapping with the file hn1roman.mbroli -------------------------------------------------------------- 3.0 Contacts: -------------------------------------------------------------- 1) Prof. GIM, Gyeongseog/KIM, Kyongsok. Internet: gimgs@asadal.cs.pusan.ac.kr http://asadal.cs.pusan.ac.kr/hangeul/ (Hangeul/Hanmal info.) http://asadal.cs.pusan.ac.kr/ (General info.) Phone: +82-51-510-2292 (office) or +82-51-947-6581 (home) Fax: +82-51-515-2208 Address: Division of Computer Science and Engineering College of Engineering, Busan National University BUSAN 609-735, South KOREA 2) Mr. Chung, Hyunsong. Internet: hchung@pitch.phon.ucl.ac.uk http://www.phon.ucl.ac.uk/home/hchung/home.htm Address: Department of Phonetics and Linguistics, University College London, London WC1E 6BT, U.K. TEL: +44 171 504 5026, FAX: +44 171 383 0752