Arpasing or vccv

Discussion in 'Q&A and Tech Support' started by Schattenfluch, Jun 9, 2017.

  1. Schattenfluch

    Schattenfluch Momo's Minion

    Messages
    10
    Likes Received
    11
    Trophy Points
    8
    Hey guys,
    I m going to record a UTAU together with my brother and was wondering which one is better?
    English vccv or arpasing?
    How long does it need to record? How understandable is it?
    I know that arpasing is a new Typ of recording and it's not finished yet.
    But which one would you recommend?
    LG Schattenfluch
    Ps: sorry for my English... I'm from Germany
     
  2.  
  3. Damien

    Damien Wannabe Worker Supporter Defender of Defoko

    Messages
    127
    Likes Received
    1,009
    Trophy Points
    108
    Arpasing.
    Could take between 30 minutes to 2 hours for a single pitch.
    Very understandable (If you, uhm, don't utterly fail at using arpasing, its not hard to get used to)

    I recommend arpasing cause it's my favorite of the two.
    You wouldn't be the first German to try Arpasing uhuhu.
     
  4. Fanicom

    Fanicom Suami's Istri Defender of Defoko

    Messages
    62
    Likes Received
    74
    Trophy Points
    33
    Definitely arpasing :sing: takes less effort to record and can still be up to par with VCCV, with the advantage of being able to generate a base oto with Moresampler for it. Saves you lots of time, eh~

    It's probably best to record if you mimic American English accent, though. I'm not a native either, so that's just an advice
     
  5. Damien

    Damien Wannabe Worker Supporter Defender of Defoko

    Messages
    127
    Likes Received
    1,009
    Trophy Points
    108
    forgot about the oto generation heh. Yeah moresampler can generate a really decent oto.ini too and you might not have to modify it that much if you're #blessed
     
  6. Sors

    Sors Immature Adorable Cookielord (´• ω •`) Defender of Defoko

    Messages
    328
    Likes Received
    341
    Trophy Points
    93
    I personally recommend both, tho it depends. While VCCV takes more work, its easy to use once the vowels are remembered, and there are a lot of usts out there. It does take a lot of time to record (bipitch took me up to 10 hours), but is easy to use and tune.

    Arpasing has its advantages, such as the short reclist, automatic oto as well as arpasing assistant, its rather hard to tune, so I'm not fond of it. You can of course do both, as to provide the best results for users.
     
  7. UTAUlikeVOCALOID

    UTAUlikeVOCALOID Ritsu's Renegades Defender of Defoko

    Messages
    98
    Likes Received
    339
    Trophy Points
    67
    Okay, so I know I haven't really tried it, but based on the sound of it when its finished and based on reading other threads, I definately think Arpasing is better. I mean its short, easy and the outcome sounds good. Besides, you got Arpasing Assistant to help you when making USTs
     
    Damien likes this.
  8. FeatheredFinch

    FeatheredFinch King Of Songbirds Defender of Defoko

    Messages
    88
    Likes Received
    160
    Trophy Points
    53
    I second this
     
    UTAUlikeVOCALOID likes this.
  9. Skiyoshi 十三

    Skiyoshi 十三 Lotus#12915 Defender of Defoko

    Messages
    250
    Likes Received
    421
    Trophy Points
    88
    For a first english bank I have to say arpasing. A lot less things to record yet it still gives about the same quality. (Personally my arpasing sounds better than my vccv.)
     
  10. partial

    partial Floating in a Fragment somewhere. Retired User Supporter Defender of Defoko

    Messages
    721
    Likes Received
    1,461
    Trophy Points
    118
    I would absolutely recommend VCCV. I stand by my opinion that VCCV is the clearest method of English singing synthesis to date.
    Pros:
    - very clear
    - easy to work with
    - USTs are a lot simpler
    - easy to manually edit USTs

    Cons:
    - takes a lot of time to record
    - takes a lot of time to oto

    Arpasing is also a great method though. It sounds about the same level as V2 English to my ears, though. Not that that's bad!
    Pros:
    - very easy to record
    - oto is automatically generated

    Cons:
    - not as clear sounding as VCCV (in my opinion)
    - USTs are very difficult to create without the arpasing assistant
    -USTs are incredibly hard to edit

    At the end of the day, it's really your decision. I would recommend VCCV, but maybe that's just because I was around when it came out and I'm old school? Haha. Maybe try arpasing to test the water, and if you feel the need to try VCCV, great. If not, also great.
     
    kimchi-tan, azazel, Arissa and 2 others like this.
  11. Kiyoteru

    Kiyoteru Local Sensei Supporter Defender of Defoko

    Messages
    2,071
    Likes Received
    2,796
    Trophy Points
    157
    A lot of people here have made great points that I agree with. I guess the only thing I can add is to correct you on something:

    It's new, in the sense that it's been around for less than a full year. However, it most certainly is a complete method. There's certainly room for improvement in things like Arpasing Assistant, but that's outside of our hands. You can confidently create an Arpasing voicebank and know it's a finished, ready-to-use English voicebank. People have made quite good work with it already!
     
  12. 수연 <Suyeon>

    수연 <Suyeon> Your friendly neighborhood koreaboo trash Supporter Defender of Defoko

    Messages
    593
    Likes Received
    640
    Trophy Points
    129
    - Ease of creation...
    Arpasing by far. I have tried VCCV (never finished); I have oto'd VCCV for folks (they make me cry inside). Arpasing is easier and faster to record. Moresampler will do the otoing for you (tho you may have to edit some of the output). The common downside to both is that they're mainly aimed at Americans and rely on having a General American accent (not necessary, but you will get comments about not being 110% understandable from trolls - an Australian friend of mine had this issue).

    - Ease of use...
    I have not extensively used either, so I can't comment on the ease of tuning. As far as the alias scheme, I find Arpabet to be the easier of the 2 to understand, where as CZ's method often makes me feel like I'm trying to translate l33t. I don't think CZ's method translates well to songs that are fast or has complex wording (a 2-3 syllable word can become 5-6 tiny notes).

    - Aural aesthetic...
    Arpasing has the potential to sound more like an actual human (to me), while VCCV sounds closer to a Vocaloid (notes can get slurred, overly precise pronunciation that can interfere with a natural flow and sound more uncanny than anything). Of course, this is all down to usage, any additions made, and ust editing.
     
    aoi・kurt, VocAddict, Tema and 2 others like this.
  13. Kiyoteru

    Kiyoteru Local Sensei Supporter Defender of Defoko

    Messages
    2,071
    Likes Received
    2,796
    Trophy Points
    157
    Future plans for Arpasing include a British specific list, which is why the 'ax' phoneme is officially supported (though not truly implemented in any current lists, to my knowledge).

    However, since Arpasing is a more open method than VCCV, you could easily try to write your own lists instead of waiting for Kanru. (Though, if you're proficient at writing your own lists, you might as well not be basing it on any standards to begin with LOL)

    I would argue that Arpasing can actually be worse with this problem. Arpasing works exclusively on diphones, that is, each note only has two phonemes. Therefore, in order to reconstruct clusters, you must have a note for every single transition. VCCV, however, does have the occasional 3-phoneme notes, such as the "str" cluster. But you can probably safely get away with not doing USTs in the "perfect/standard" method, and omitting transition notes for the sake of clarity.

    Absolutely. They're both lists of the same language. What you put in is what you get out. For example, Delta's CVVC list may seem like it lends voicebanks a Japanese accent. However, that's because most users of that list have Japanese accents. When a native speaker records it, it sounds like the accent of a native speaker.

    Though, one of the concerns to be had with VCCV is that the default BGM involves recording at 120bpm. (This was for the sake of easy calculations in OTO values. It's not fun trying to do math with weirder tempos!) At times, it may be too fast to record with a singer-ly tone. With Arpasing, you have the freedom to record at whatever speed you like, as Moresampler is analysis-based and not strictly based on regular timing. So, if you want to take more care in your pronunciations, you may want to turn BGM off.

    Good luck, and I hope you figure out which list works best for you.

    EDIT: I forgot to mention something! If you're recording voicebanks with someone who doesn't use UTAU, then you'll definitely want to choose a shorter list to keep their interest (especially if it's an easily bored child).
     
  14. partial

    partial Floating in a Fragment somewhere. Retired User Supporter Defender of Defoko

    Messages
    721
    Likes Received
    1,461
    Trophy Points
    118
    I definitely agree with this!
     
  15. Schattenfluch

    Schattenfluch Momo's Minion

    Messages
    10
    Likes Received
    11
    Trophy Points
    8
    Thanks, those are great tips.
    I think I'll go with Arpasing since the time I have to record is really limited and like you have pointed out, my brothers attention span is very short...
    Thank you
     
    partial and Kiyoteru like this.

Share This Page