It is quite a lengthy list, I'd recommend looking into stringed utterance to lower the amount of files needed
Also, as the creator of the reclist, usually it's one's own responsibility to create a base oto.ini for the list, n have others learn or reference it. Just my two cents!
It is quite a lengthy list, I'd recommend looking into stringed utterance to lower the amount of files needed
Also, as the creator of the reclist, usually it's one's own responsibility to create a base oto.ini for the list, n have others learn or reference it. Just my two cents!
Is there anything else I have to do before updating the list??
And also, I thought it was optional to add the base oto.ini
Do you want to voice it or know anyone who might want to voice the list demo??
Sorry but I can't tecord rn
Because my windows machine is not accesible right
How does stringed utterance work??
Hello! I'd recommend making sure you type out everything at once that you'd like to say before posting a reply, it stops the thread from getting gunked up abit!
1. It's okay to not be able to record yourself, but you can still type an oto.ini for a voicebank without any wav files :D You'd have to type in the name of every wavfile that's supposed to exist, and make some base parameters for each set of those files (Ex, how CV would have a different base parameter than VCs)
A really good tool I use for this is called Oto Adjuster (made by Crystalex93). You can select the first mora (syllable in a reclist) and apply a parameter to just that set, or the opposite!
2. While it most definitely is optional, it's highly recommended for custom lists, especially for English lists. Not everyone is going to know how it is supposed to be configured, so it's extremely helpful for the learning curve!
Unfortunately, I use my own English recording list and I'm not particularly interested in recording for other lists, unless they are below 289 files + fully functioning English.
3. You don't have to record it yourself, there's no pressure! I totally understand that kind of situation
4. Stringed utterance has multiple mora (aka syllables that are sung). This is common in modern CVVC and VCV voicebanks. You use the term Mora to identify how many syllables a string will have in your recording list, and this recording style greatly reduces the amount of time needed to record a voicebank. For example, a 7-mora recording list would have 7 syllables, like this ; "ka-ka-ki-ka-ku-ke-ka"
You still get your CV sounds + Your VCs, but in a single line of recording. It's a bit more tedious for English, but the payout is definitely worth it!
Not to self-advertise or anything, but I do have a Google Spreadsheet full of information about English phonemes, if you'd want it. For now, though, I'll link the ARPAsing phoneme guide for reference
First... Please just type everything you need to say into one post to avoid a lot of unnecessary merging. Second... I'm an adult. I was at work. Many people who frequent (or used to frequent) this forum are adults. Many people have jobs, school/college, or both. It's not respectful to bombard others with requests for conversation, whether they are available or not. Thirdly... This is a forum. Forums do not operate like discord or other chat services. You need to have patience and wait for responses and you will be notified when a response to your thread is given. Don't expect other people to instantly reply because that's not how forums generally operate.
Now...
If you're going to make a reclist for a given language for someone else to use (and you're not going to use it yourself), you want to include the following...
1. Pronunciation guide: can be a word with the target sound, a link to a sound guide, or a wiki for the sound in IPA.
2. A table that represents your symbols and a comparison chart if necessary (since this is for English, the charts you'll want to use as a comparison guide are Delta, ARPAsing, Vocaloid's X-SAMPA, and CZ - someone familiar with these forms of English reclists will be able to cross-reference this chart in addition to your pronunciation guide when they are unsure of what sound they need to make for your reclist). This is only really needed for languages that can have variations in aliasing methodology (for example: English, Korean)
3. A base oto for ease of configuration. You don't need anything special to make a base oto. All you need is a notepad editor (you can save as .txt and change the filetype to .ini when you have access to a computer).
name_of_file.wav=alias,0,0,0,0,0
4. a test run demo of the recorded and configured bank to give an idea of what it may sound like. While this isn't completely necessary, it will make your reclist more appealing to use since... you have confidence that it's actually usable. No one wants to record a bank - especially one as large as English - using a reclist only to find that it's missing necessary sounds (if it's missing sounds on purpose, then you can state the workarounds to get a particular sound combination to work).
Ok so you can make consonant blends (not ones like th,ch and, zh/sh which I already added and are not just pronounced by having a t before h or a h after c
How to make up for missing sounds
Kw sounds
Conquest- Ko ong k/ /w we es s/ /t
Just add a schwa inbetween 2 notes which contain the consonant you are trying to blend
I can (and will) add a base oto.ini but it will be blank
The pronounciation guide comes with the reclist
And it was late at night when I replied and i had a looong Eid vacation (i am not a muslim)
And I do have another computer but it's a mac
I hope anyone can voice this for me
Examples:
th// k/ /l la ak s/ /t ta ap p/ / d ti ik ki ing fo or re ev ve //r //g go
the clock stopped ticking forever ago
ha aw lo ong ha av I bi in //p I dO On n/ /t nO
how long have I been up, I don't know
I ke en n/ /t ge et // g/ /r ri /p b// //t I ke en n/ /t le et gO
I can't let a grip but I can't let go
the er wa se en n/ /t en ni thi ing to ho ol l/ /d on n/ /t to thO
there wasn't anything to hold onto though
(Just so you know this reclist is for people who want something a step from normal japanese CV)
Also compatible with japanese
do od da at te i ko ot to o us so ot te ha i it te mo od do or re en na i
Do datte i kotoo, usote haite modorenai
ji ik ko on na an n/ /te ya at te ko on nai
Jiko nante yatte konai
ub at ta yo on ni ub ba aw wa ar re et te
Ubatta yoni ubawarete
ki iy yo da at te ka ma aw wa an na i
Kyou datte kanawanai
Could we ask again why you're unable to test out your own reclist?
You don't need to make an extremely high quality voicebank, even recording with your built-in microphone would be alright. It's fine and dandy to come up with plans for how a reclist might work, but I really think you ought to record it yourself and try using it in UTAU before asking others to give it a go.
I only have access to a Mac right now, and although there is Oremo (and Akorin which you told me about) for mac I can't get those to work
and my parents won't let me install anything because macs are expensive
(they allow me to install voicebanks tho)
Do you mean recording CVC syllables like "bab" "beb" "bib" etc.? That's generally how English CVVC works in the first place.
I looked at your list, and right now you've got your CVs and VCs separated. Also, you should know that you can't use characters like [/] in file names, at least not on PC.
You've got the general idea, but this reclist has a lot of common mistakes, like equating the written alphabet to phonetic sounds. In reality, verbal English is much more dense with sounds than our alphabet implies. I highly recommend looking into English phonology and phonetic transcription systems like X-SAMPA, as well as familiarizing yourself with more complex reclist types.
So I changed the / to 1
Schwa list
1 is a schwa but more like a
2 is a schwa but more like e
3 is a schwa but more like i
4 is a schwa but more like o
5 is a schwa but more like u
I'm on a mac so i forgot that filenames can't have slashes, Also i removed E (i) and U (yu) since there are workarounds
There are alot of workarounds using this list, so it might be super choppy
MediaFire is a simple to use free service that lets you put all your photos, documents, music, and video in a single place so you can access them anywhere and share them everywhere.
Unfortunately those sounds are still pretty necessary for English. If you want your voicebank to be understandable, you'll want it to be able to say words like "the" and "this" with a voiced fricative and "thing" and "think" with a voiceless one, which are all very common words.
As far as vowels, schwa only refers to /ə/ (the "a" in "about"). It's a very lax central vowel that sometimes sounds more or less like /ʌ/ or /ʊ/ ("u" in dug and "ou" in "could"). It's also not very useful to think of them in terms of the "a e i o u" since they don't graph 1:1 with the written language. Assuming an American accent, generally you have all of these vowels at least: /ɑ/ /ɛ/ /i/ /ɪ/ /u/ /ʊ/ /ʌ/ /ə/ /æ/ /aɪ/ /eɪ/ /ɔɪ/ /aʊ/ /oʊ/ and /ɚ/
It's technically possible to get away with fewer vowels, like what people often do to make Japanese voicebanks sing in English, but it won't be very easy to understand. I definitely get the want to have as small of an English reclist as possible, but it's just not very practical; English is phonetically complicated.
One final note, I'm not sure why you have "x" /ks/ as it's own consonant? You don't include any other consonant clusters besides /ts/, and /ks/ only occurs at the end of words.
I encourage you to keep going, but you really should do more research before you try to find shortcuts. Better to start too big and pair down than to start too small and be missing key sounds.
![[K] of Honor](/data/medal/1_1406930084l.jpg)
