Synthesizer V fails to import MIDI output of MuseScore, in terms of importing Lyrics and Dynamics, automatically adjusting Time Signature (i.e., 4/4, 3/4, etc.) changes. It had been explained that the reason is that how to include Lyrics in MIDI is not defined in MIDI standard, for example. However, knowing how Lyrics (and other elements, like Dynamics and Time Signatures) are included in MuseScore produced MIDI, it was demonstrated tat it is possible to convert the .mscz file automatically to .s5p by using a Python script. (MuseScore to SynthV - YouTube). It means that it may be possible to equip Synthesizer V to directly import .mscz file into Synthesizer V. Since the script in the above link seems to be for Japanese only, I am more interested in a feature for automatic language detection for Lyrics import. Also, compared to Vocaloid, Synthesizer V AIs are limited. I hope that the future AIs can handle more languages or more AIs that can handle other languages other than English, Japanese, Chinese.
I don’t know about languages other than English, but I have experimented with MIDI and Synth-V with a few different programs. It is not a failure of Synth-V if the MIDI file does not have lyrics, but rather a failure of the program - in this case Musescore - that did not embed lyrics in the file. If the MIDI file has lyrics embedded Synth-V will import them correctly. It is optional whether a MIDI file contains lyrics, so Sibelius and Encore files do contain them and open in Synth-V with lyrics, but Dorico does not, and only opens with notes. I have notified Dorico, and they will add this feature in the future. You should advise Musescore that the MIDI export should include lyrics, and perhaps they can include this in an update as Dorico is now doing.
In any case, both Musescore and Dorico have XML export (which always contains lyrics) and I have tried using online XML to MIDI converters with Dorico files and can get Synth-V to receive lyrics from Dorico by doing this. I suspect you can do the same with Musescore: export the file from Musescore as XML and convert it to MIDI with an online converter, then open that in Synth-V and see if it works. If one converter doesn’t work, try another (search on the net) and you may find one that works.
Synth-V will always get the tempo information - including changes - from a MIDI file, but only if you import it and select “open in a new file.”
MIDI and XML files with lyrics divide multi-syllable words with a dash (-) between the syllables, whereas Synth-V reads the whole word and needs a plus sign (+) after the word for each syllable after the first. So MIDI or XML will write “goodbye” as “good - bye” but XML wants it to be “goodbye +” It doesn’t matter how many syllables in the word, just put as many +'s as necessary (like, say “telephone + +” or “interesting + + +” Also, XML files will add a dash “-” to connect two or more notes singing the same syllable (called melisma) and happily Synth-V reads and understands this.
So even if you get lyrics into Synth-V from Musescore, you will still have to fix all the multi-syllable words. Sometimes they will still sound right with them being broken up, but a lot of times it is pronounced wrong. I find the best way to fix these is to copy all the lyrics (either using Claire’s script or CTRL-L and manually copying them) and paste them in a text editor to edit them carefully, and then paste them back in Synth-V with CTRL-L or Claire’s script.
I have been working on this, and am now experimenting with offline text converters and programming them to automatically convert MIDI / XML lyrics to Synth-V format, and also to add + to words in a text file (like from the the internet) and will post these convertors when I have them perfected.
Hi. Thanks for the info. I will ask MuseScore whether it can improve the MIDI format so that it may be compatible with Synthesizer V style, and will see what happens. I will also try to use XML and see whether it works. Given the non-existence of the compatibility at the moment and that XML-method still would need a post-import tweak, however, I am rather reading Lyrics off the original music score direct and entering manually to Synthesizer V tracks by Cmd-L (as a wholesale). I select all the notes in a track (or a portion of), and then press Cmd-L. Then, I enter the whole Lyrics (or of the part). I observe the notes and Syllables, then, I can tell the syllable-assignments to notes (I enter them with “-” where necessary, although it does not seem that Synthesizer V cares much whether they are there or not). After making sure that assignment is correct, I use Cmd-L again to import the whole Lyrics out to a Text tile. Then, I tweak it for minor some pronunciation corrections or improvement, as I use English AI for Latin, etc. In terms of the conversion script that I mentioned, it seems it can do only Japanese Lyrics, since it is probably trying to produce correct phonemes for Japanese Lyrics, rather than assigning Japanese syllables directly to notes in Synthesizer V. As to the problem with Time Signature, one of the problems stems from Synthesizer V’s behaviour that does not allow the score windows to be closed then let user to chose “New Score” from the File menu. If it is allowed, one can close the current score and then go to the file menu “New Score” to start a fresh doc. Currently, if one closes the main (and the sole) window, the app quits. This is not a normal behaviour. Yes, one can still choose “New Score” from the File menu when the main window is still open, then Synthesizer V asks whether the use wants to just close the current window or to save it, but this is the only behaviour at the moment. Nothing is wrong with it, but this is not what a user normally do. Usually, the normal psychology, of the user who wants to start a new work, is to close the current window and then select “New Score” from the File menu.
Yes, Yoichi, the method you use (Cmd-L) to bulk enter lyrics, or to copy them and edit them is a great time-saver. You can go back and forth, copying, editing, and re-doing them with this method, and it is a quick way to learn how Synth-V responds to different words and spellings.
Using an offline text editor is also handy, which is what I do. I should mention that using Claire’s scripts called “copy lyrics” and “paste lyrics” (each assigned to a single key) is quite fast for this purpose, because all you have to do is select the phrase or whole song and hit one key (for "copy lyrics’) and then in your text editor hit paste and you have the lyrics ready to edit. When you are finished, copy the lyrics in the text editor, and if you have not changed the note selection in Synth-V just hit the other key (for “paste lyrics”) and it will re-write them back to Synth-V. If you keep your text editor open at all times, this is faster than Cmd-L because you don’t have to select the notes, then Cmd-L, then select them inside Cmd-L, then paste into your text editor, so it turns three key-strokes into one.
If you start using this method, you can then extend it to phonemes, using Claire’s scripts, “copy phoneme” and “paste phoneme” because this gives you the ability to bulk copy or edit phonemes the same way you do with English words with Cmd-L. Really, Synth-V should have the equivalent of Cmd-L for phonemes, which has been suggested by users as an improvement in the future. But for now, Claire’s script can do this.
Once I started using this method, with either words or phonemes, I starting writing different search and replace macros, and/or autocorrect files in Microsoft Word, to quickly convert whole lyric files from one format to the way Synth-V likes them. I am still working on them, but once they are perfected I assign to one key in Microsoft word, and they can really speed things up. You can have whole autocorrect dictionaries in Microsoft Word, like, say, to convert American accent to British accent, and apply that in bulk to the phonemes, (not the words) and it works great. I just have to keep adding phoneme changes to the autocorrect dictionary in Word, but with time this dictionary would be very useful, and could do any accent at the push of a button. There are lots of other things you can do to auto-convert, hyphenate, or de-hyphenate, add “+” after multi-syllable words, and things like that, which affect the whole lyric file with one key-click in Microsoft Word.
Hi. Thanks for that. I usually leave the external Text Editor open and leave the Cmd-L window open, too. Where can I get "“Claire’s scripts called “copy lyrics” and “paste lyrics””, by the way? It seems it is a better way to do it. And also, I am interested in ““copy phoneme” and “paste phoneme””. As I deal with German and Latin Lyrics, it may be easier to tweak English phonemes than trying to come up with pseudo English Lyrics (that use combinations of English alphabets, such as tzche, ee, oo, etc. which mimic the pronunciation of other languages somewhat less satisfactory). I assume that that uses English phonemes. Since I guess that the phonemes for English does not cover all of IPA (International Phonetic Alphabet), I guess there are some limitations.
You can download my scripts from this webpage. Keep in mind scripts require the Pro edition of SynthV Studio.
Clicking the “.zip” folder in the upper right will let you download them all at once instead of one-by-one.
Thanks for the link!
It’s clear that navigating different music and lyric file formats is a challenge. Have you considered taking inspiration from image converters, like the one found on Converter365 Image Converter?
It supports a multitude of image formats, requires no installation, and prioritizes user security. Applying these features to a music file converter could be highly beneficial.
I’ve written a MuseScore plugin mscz2svp to export from MuseScore to Synthesizer V Studio. It only works on MuseScore 3.6.2 right now, though, since that’s the version I mainly use and I’m waiting for the plugin API to stabilize on MuseScore 4.
That’s great, and I can’t wait to see one for MuseScore 4!