How do you make a English voicebank?

Discussion in 'Help and Advice' started by Wenyu, Aug 10, 2017.

  1. Wenyu

    Wenyu Teto's Territory

    Messages
    35
    Likes Received
    29
    Trophy Points
    23
    I mean, how would you OTO it and stuff?
    And it's CVVC or whatever right? How do you oto that? And i looked for a reclist and theres an "unstringed" and "stringed" reclist. What does that mean and does it make a difference in oto'ing it? And is creating a power scale voicebank any different for english? And i've never used an english UTAU before so how would it's aliases look like? The reclist is really weird, i see stuff like chax, uwth, things i wouldn't normally write in english, and how do you pronounce that???? and on some reclists i saw something like
    _Q8Q&n
    _Q8Q&n


    also, if you want, please give tips on how to record a good english voicebank, like maybe pronunciation wise? I hear a lot of voice banks where the r's sound weird and stuff and its really nasally and the voice just sounds really mushed and unclear and not powerful.
     
  2.  
  3. Kiyoteru

    Kiyoteru Local Sensei Supporter Defender of Defoko

    Messages
    2,071
    Likes Received
    2,794
    Trophy Points
    157
    I've already linked to some reclists here in this thread.
    http://utaforum.net/threads/looking-for-english-vb-reciclists-c.16477/
    Those CVVC Stringed and Unstringed reclists are an older version of PaintedCZ's work, so if you want to use the same style of phonemes, consider using the more current VCCV reclist.
    All three (VCCV, Delta, Arpasing) have detailed OTO guides. All three have resources regarding the pronunciation of phonemes.

    Yes, recording a powerscale bank would work the same way as with any other bank. Delta's list shouldn't need any other adjustments. However, with VCCV, you will want to combine all the subfolders and sublists into a single folder and list. With Arpasing, there are notes in the tutorials regarding how to handle multipitch.

    I highly recommend recording with a singing tone of voice. English pronunciation varies between speaking and singing, so a singing tone will give you a more natural sound when recreating singing vocals in the end.

    If I recall correctly, you're a Mac user. If you have any questions regarding how to set up things like OREMO, etc. on Mac, feel free to ask me.
     
  4. 幸兔雪 (Yukito Yuki)

    幸兔雪 (Yukito Yuki) Defoko's Slaves Supporter Defender of Defoko

    Messages
    516
    Likes Received
    1,335
    Trophy Points
    123
    That link what Kiyoteru (previous person) linked you, when you scroll down you can find phonetic table made by me which sound is which. For example [æ] sound like in English word cat is encoded as @ in VCCV, { in Delta and in Arpasing as ae.

    The reason why we use weird "symbols" to write English phonetic because unfortunately English isn't language that "write how they pronounce it".

    "a" in "bake" "about" "bat" "cat" are all different sound! So, to know which "a" sound it is it's a must to use an encoding. A "pronouncing code language" for each English phoneme.
     
    kimchi-tan and Kiyoteru like this.

Share This Page