Hello,
i’m using Cubase 15 Elements on Windows 10. Problem with Omnivocal beta:
I’m recording a melody with my MIDI controller keyboard. It’s later played back correctly and at the recorded pitch.
However, when I try to enter lyrics, it doesn’t work:
90% of the lyrics end up on the first MIDI note, the rest on the second MIDI note of the MIDI track.
Example: I’ve recorded 10 MIDI notes.
I have 10 words, 9 of which end up on MIDI note 1 and 1 word on the following second MIDI note.
The remaining 8 MIDI notes are empty; Omnivocal just “sings” “Ah-Ah-Ah…”
Even correcting the pronunciation doesn’t help because Omnivocal either “sings” it completely wrong or not at all.
What could be the problem?
Is the melody recorded at too high/too low a pitch?
Do I need to select a specific instrument on the MIDI controller for recording (i played a free Steinberg piano instrument)?
… Unfortunately, I can’t find anything about that problem online. Any hints?
Thank you 
Can you make a screenshot from the Key Editor with all notes visible and the first one with the text selected. I’d like to see the Info Line of the editor, too.
Make sure when you enter the lyrics you only select the first of the 10 notes, not the entire 10 notes. This feels like a UI weirdness – i.e. counterintuitive – but, if you only select the first note, and type in the 10 syllables in the text field on that one, it should correctly apportion them across the 10 notes.
1 Like
Yes, i selected ONLY the 1. note to enter the lyrics …. ALL lyrics went into the FIRST note.
I made 2 screenshots, so hopefully you get all the info
Problem still unclear…?
…or does nobody have any idea what the problem could be? 
Maybe try to break up the lyrics in several lines and paste the text of each line onto the relevant notes.
It is unclear.
It is normal for all the lyrics you enter with the first note selected to show up there (but there should only be the first phonetic syllable in square brackets at the end of that textual string). However, each subsequent note should show the 2nd phonetic syllable, the third, and so forth. But, in your case, if you literally entered the lyrical string you show in the text box with only that first note selected, then the second note should show the phonetic representation of “thing’s” (the first note’s phonetic representation should show that for “some”, and the third should show the representation for “fol”, and so on). But your second and third notes appear to have the phonetic representations for “you” and “and”, respectively, which is skipping syllables.
I do recall that when I was initially thinking I had to select all the notes for a lyrical phrase being entered there were results that didn’t seem to make sense, which is why I asked about that earlier.
The only other thing I can think of here is the question as to whether you pasted the lyrics as you have shown them (i.e. with carriage returns/new lines) on that first note, for example from a text editor or word processor. If so, maybe there is something weird in its behavior in that case. Beyond that, I’m clueless on potential explanations.
What I would suggest however, is that you break the lyrics up into smaller phrases, for example just putting “something’s follow you home” (without the quotes, of course) while only selecting that first note, and see if that distributes things correctly. In general, when putting in a phrase, versus a word or syllable at a time, the way Cubase displays the lyrics gets a bit confusing/unwieldy due to the long string in the first now, where you have to scroll horizontally in the text box to see the phonetic representation if you want to tweak that. The longer the textual line is, the harder it is to deal with that, and, also, if things get broken up in a way other than you intended, for example due to melismas and not putting in the right number of hyphens, the more work will be needed to fix the syllable allocation. It’s too much of a hassle to just go word by word, but I find doing it approximately one vocal phrase (i.e. between breaths) at a time is a good middle ground for the tradeoff between efficiency in entering the initial lyric and efficiency of making any needed tweaks.
Thanks, I’ll give that a try. It’s 4 lines of text, I’ll enter each line individually.
Exactly, the first note contains all the lyrics, and from the second note onward, only individual letters, etc., are displayed.
Yes, it was entered as plain text (notepad.exe) without any formatting characters or line breaks. The four-line display was only to show the text length.
I’ll try entering it line by line. Let’s see if anything changes…
Thank you for your contributions so far. I’ll experiment with it some more.
Since no one has mentioned that a specific input method should or must be used, I assume that a regular piano at a normal pitch will suffice.
Strange new thing… I’ve noticed something else: some of the lyrics (that I can identify) sounds slightly Japanese-accented… as if, for example, Photoshop were installed in Japanese instead of English. But there’s nothing you can do about it except install the plugin. I didn’t specify anywhere that I wanted it in Japanese…
So, I’ll keep trying… 
Omnivocal supports two languages, English and Japanese. It tries to detect automatically which language is used when entering lyrics. Maybe there are some syllables which are are common in both languages. Also, others have reported that they noticed a foreign accent when using English language.
Okay, I was able to fix the problems 
Three things you really need to pay attention to:
1.) There must be exactly as many MIDI notes as there are syllables in the text/lyrics. It’s not the number of words, but the number of syllables in the words that’s crucial for the text to be automatically distributed across multiple MIDI notes with the correct phonetic transcription.
And you shouldn’t enter too much text at once… better to enter two sets of four words than one set of eight.
2.) Some English words aren’t recognized correctly. Not a big problem at all. So, the lyrics should be kept simple. If a word isn’t recognized at all or is poorly “sung,” try finding an alternative. That helps too.
3.) The problem with unclear “singing” or almost Japanese pronunciation apparently arose because some MIDI notes were duplicated. This happens quite easily when editing with the mouse. Once the duplicate MIDI note is deleted, it works again.
=> These are the three things I now pay attention to.
Now it works as it should—and it’s really fun.
The result is fantastic! 
2 Likes
Yes, english + japanese - I know.
Well, the accent was due to MIDI notes being doubled on top of each other…
