search

Sleepwalking / Rocaloid-old

OBSOLETED! Moved to http://github.com/Rocaloid
GNU General Public License v3.0
63 stars 16 forks source link

音标记号问题 Phonetic symbol issue #2

Closed m13253 closed 11 years ago

m13253 commented 11 years ago

看了你的字典，发现使用的是你们自定义的发音记号，建议使用国际标准的 X-SAMPA 记号。 VOCALOID 的发音记号就是 X-SAMPA 记号。 然而也有例外，如洛天依音源因为某些 bug 导致音标和发音不完全对应。（如拼音 bo 发音应为 p uo 实为 p o 或者拼音 er 洛天依的发音是 Ar 而音标写成了 `@``。） 我们先不吐槽洛天依因为赶工时导致的各种 bug，建议 Rocaloid 使用国际标准的 X-SAMPA 音标格式。 如果你愿意，我可以提供拼音到标准 X-SAMPA 的转换表。 m13253/pinyin2xsampa。（你可能想使用修改一点的 X-SAMPA 便于实现） 我也愿意参与 Linux 移植计划。（当然是等到 C++ 重写之后）

I have had a glance at your dictionary and have found that you are using custom phonetic symbols. I suggest you use X-SAMPA phonetic symbols which is International standard. In fact, the phonetic symbols that VOCALOID uses are X-SAMPA. However there are exceptions. For example, some bugs of Luo Tianyi soundfont resulted in inconsistency between phonetic symbols and the actual pronunciation. (Such as bo in Pinyin should be pronounced as p uo instead of p o, and er in Pinyin should be Ar instead of `@``.) Despite those bugs found in Luo TY soundfont due to terrible work quality before the deadline, I recommend Rocaloid use X-SAMPA phonetic symbols. If you are willing to, I can provide a table converting from Pinyin to standard X-SAMPA. m13253/pinyin2xsampa (You may want to use a slightly modified version of X-SAMPA for easier implementation) I would like to participate in porting to Linux as well. (Of course not until it is rewritten with C++)

Sleepwalking commented 11 years ago 
    感谢你提供转换器！我也认为转换成X-SAMPA会更科学。但是目前有如下几个原因使我们无法应用X-SAMPA：
  1. Rocaloid的初音音源库是从Vocaloid合成得来的，但是初音无法发出所有X-SAMPA的音标。Rocaloid符号系统甚至都有一些空缺。
  2. X-SAMPA无法准确描述音素间过渡。而Rocaloid发音符号和CVS结合紧密。
  3. X-SAMPA的辅-元组合数量太过庞大，完整音源库制作可能耗费数年。 我觉得折中的办法是改进Rocaloid符号系统，使其更接近X-SAMPA，但是完全应用X-SAMPA不大可能，如果技术上没什么突破的话。

【顺便求教issue tracker文字格式怎么搞……我刚开始用GitHub

Thank you for providing the converter! I also believe it would be more scientific if we use X-SAMPA. But currently we have several problems that X-SAMPA cannot be applied:
  1. Miku's database of Rocaloid is synthesized from Vocaloid, and Vocaloid Miku is not able to pronounce all X-SAMPA soundmarks. Even the Rocaloid Symbol system has some vacancies...
  2. X-SAMPA is hard to describe the transition between phonemes, while Rocaloid Symbol fits well with CVS.
  3. X-SAMPA has too many Consonant-Vowel diphones, building such huge sound database may cost us several years... I think the better solution is to improve the Rocaloid Symbol system, to make it some similar as X-SAMPA. However it's not possible to replace it with X-SAMPA, if there's no significant improvement in technology.

【By the way, could you teach me how to apply format in issue tracker... I'm new to use GitHub

m13253 commented 11 years ago

初音ミク的发音记号就是X-SAMPA……

Sleepwalking commented 11 years ago

那只是个非常非常不完整的X-SAMPA而已…… But that's just a very incomplete X-SAMPA...

Vowels a i M e o Nasals n J m N m' N' N\ VoicedPlosives g g' d d' b b' VoicedFricatives z Z h\ VoicedAffricates dz dZ Liquids 4 4' Semivowel j w UnvoicedPlosives k k' t t' p p' UnvoicedFricatives s S h C p\ p\' UnvoicedAffricates ts tS GlottalPlosive 无 Breathings in out br5 br4 br3 br2 br1 Silence Sil Asp

m13253 commented 11 years ago

引用：

顺便求教issue tracker文字格式怎么搞……我刚开始用GitHub

文字格式是 Markdown，可以 Google 一下自学。 而且源里面 .md 后缀的文件也会 Markdown 解析。 比如如果你想写一个带格式的 README，就放一个 README.md 文件，否则就放一个纯文本的 README。

还有关于到底是用中文还是英文的问题，如果你希望自己的项目被外国人看懂，就用英文；否则用大家都能看懂的中文也行啊。