Playing back - delay on vocal lines.
A while ago I reported on several playback issues; some of these were accepted, some were proved to be non-events and some were left in limbo. It was a mistake to submit these all at once. In general I'm managing to live with MS4 playback via various workarounds and "getting used to it".
One or two issues I do still think are errors. Here is one.
In the enclosed piece, the voices are often subject to a delayed attack. This is much more evident on the on-line playback than from within MS4 itself - you can hear the difference.
I have enclosed the link to my piece on musescore.com, https://musescore.com/user/1600516/scores/12407080
and attached the local file below.
For example:
In the musescore.com on-line version: Bar 13, where the sopranos and tenors sing a triplet, twice, on each occasion, the triplet starts slightly late, the effect being more like a quaver rest followed by three normal quavers, instead of a true triplet. The same thing happens in bar 10.
There are numerous other examples in the score where a sensitive ear will pick out delays in the vocal production, but I thought I would stick to these to make matters simple.
In bar 30 the soprano quaver at the end of the bar is delayed by approximately a semiquaver.
In 34, the soprano is fractionally late throughout the bar.
You may say that these nuances in playback are trivial, but I use the on-line presentation to share my pieces with potential performers, and it does detract from the overall reception.
As you see on F10 I have Audio effect set to no effect, and the reverb dial on Aux Send at about 10 o'clock. What ever that does!
Thankyou kindly.
Ali W.
| Attachment | Size |
|---|---|
| Grave Goods 184.mscz | 283.13 KB |
Comments
I concur that vocal triplet playback is not always accurate. And worse in the online version.
No one here on MuseScore.org can do anything about playback on the .com site.
I think what makes the problem stand out more (online) in measure 13 is that there is no distinct down beat. Not like measure 10 where most of the voices start on beat one. In measure 13, other voices are holding and cut off on the and of one. If you are counting, the late start of the soprano stands out. Unfortunately, we can't isolate the soprano line and listen to see how accurate it is. This the nature of playback and mixing of recorded sounds. They just don't always work very good.
But even in your score, playback of many of the triplets is uneven. I changed the voice to clarinet, and it seemed better.
I suspect as general improvements are made to Muse Sounds, some of these problems will go away.
Have you tried the basic vocal fonts. They don't always sound as good. Improvements can be made by applying some of the EQ effects.
In reply to I concur that vocal triplet… by bobjp
Thanks bobjp. I suppose I should report this on the musescore.com site, but surely if I encounter similar problems on the local playback AND the com site - it must be worth considering whether the deficiencies stem locally (i.e. from within my program) and point to a weakness in something that is not musescore.com?
It would indeed simplify the situation to go back to the basic sounds, but I'd be the first one to admit that the Muse Sounds are infinitely better (in most respects) than the basic ones.
I have lived with Musescore 4 for a while now, and its evident to me, at least, that the sound quality with Muse Sounds is much improved on both instrumental and choral lines. Especially with headphones, the tone quality is fine.
However, one annoying thing still persists, and that is the delayed attack on voice lines.
Headphones / speakers - makes no odds. I can't distribute my choral pieces to people without apologising for the indistinct attack on each vocal line. And I cannot believe that this is due to any facet of my system because I don't get the delay on any instrumental parts.
The attached piece (She Walks) illustrates the delay especially on quicker moving lines. Do you agree?
In reply to I have lived with Musescore… by Ali Wood
I panned your score and turned off reverb.
In reply to I panned your score and… by bobjp
Perhaps a little better, but if you listen closely you will hear the unevenness in the placement of the quavers. This is evident throughout the whole score. It is only marginal, but all you have to do is listen to a version without Musesounds to realise what the difference in precision is.
I get this with ALL my vocal and choral scores. Perhaps I'm over-sensitive, (and I've been banging on about this ever since Musesounds appeared), but anyone conducting a real choir who sang with this level of delay would be exasperated. My friends notice it too.
It is particularly evident in faster sections (as one might expect if the delay is constant) and in rhythmic pieces.
OK, sometimes you will be listening through a piece and get used to it, and only notice it at the start of passages, But I maintain that the fault is there, and that it is a shame that the sampling can't get a choir to start singing exactly on the beat like it can get the sound of a violin, for example, to start precisely.
I have sent previous samples to this forum before, and the reaction from you good people is varied, some recognise it more than others.
BTW, reverb is all about the continuation of a note, not the delay on the start of a note, isn't it?
Thanks.
In reply to Perhaps a little better, but… by Ali Wood
Yes, reverb is about what happens to the end of a note. But think about what that might do to the beginning of the next note.
I am not a cheerleader for Muse sounds. Except to say that for what I do, they are better than most anything I have used. That includes Sibelius. Muse vocal sounds are indeed a bit sloppy. As are many of the sounds. People complain all the time about lack of precision in fast passages. I get it. But think about using a different sound for the music you send to your choir members. Maybe piano. In a rehearsal, the director might bang out a part on the piano. Many have posted on the forums that they send music to their choir members that uses woodwinds.
And consider that the human voice, in general, can't attack the start of a note the same way a trumpet would. Yes, made worse by Muse vocal sounds.
None of us know this piece like you do. It's a wonderful piece, BTW. Yet there are a lot of different things going on at once. Which is part of what makes this piece. Playback must be as accurate as possible. This is why I made those two suggestions. Playback reverb is artificial. In your score everything is panned to the center. I don't think any director would have his choir all stacked on top of each other.
I opened your score in Sibelius. By default the notes are separated. Not very good as you have the instruction about always legato at he beginning.
Consider that sound fonts do not blend together the same way live performers do.
You might consider experimenting with the new Staff Text sound flags. There are some things that might clean up the vocals.
In reply to Yes, reverb is about what… by bobjp
Thanks for devoting so much thought to this. I accept a lot of what you are saying.
You acknowledge the deficiencies in Muse Sounds - my question is, if so many of the sound fonts and associated playback is good, why are some less good than others?
The leader of another (more prestigious) choir that I sing in, indeed assesses the merit or otherwise of a new piece by playing it through on the keyboard (he is a very good pianist). Like many other musicians he is not impressed by the sound produced by Musescore, Sibelius or Soundcloud and prefers to judge for himself. Not that a piano playthrough is necessarily an indicator of the quality of a work! (eg Whitacre's Lux Aeternum)
I opened the piece with panning spread out like you suggested. It doesn't get rid of the lag, but it does improve the clarity of the parts, so thankyou for that.
I have tried the metronome, but if you listen really closely, you'll realise that the metronome hides, rather than disproves, the discrepancy.
I didn't really understand your point regarding the Sibelius note separation?
Many thanks again. A very interesting discussion. Many composers look down on the idea of playing back one's music, but there was a very interesting chat on the BBC about this recently. Lots of young composers are submitting to (and doing well in) competitions with entries that have been composed with a great deal of help from playback.
Conclusion: I need to think the music through in my head!!
In reply to Thanks for devoting so much… by Ali Wood
Ali,
Muse violins aren't very clear in fast passages. The trumpet sound can be sloppy and thin. Among many other problems.
You choir director can asses choir music on the piano. But how about orchestra music. I know he is only interested in choir music. But there is much more to music than pitch and duration. So much more. Consider that sound fonts, while recordings of real instruments, can't blend the same way real players do. I have performed in large orchestras playing well known works. Believe me, there is nothing that comes close being surrounded by the music coming to life all around you. You hear first hand how different instruments sound and blend together, up close.
Notation software has never been intended to be the final say in what a piece should sound like. But only a reference point. Much like your director's piano playthrough. So, sure, practicing musicians don't like playback. Nor do they have to. It's not a finished product.
What I hear in your piece is not so much a delay as just the font's inability to render the music cleanly. There is a font you might try. KBH-Real-Choir-V2.5. I think I got it from a site called Soundfonts 4U. Or something like that.
My guess that composers who look down on playback have the luxury of having their works performed live. I get it. It must be nice.
In reply to Yes, reverb is about what… by bobjp
> "think about using a different sound for the music you send to your choir members. Maybe piano. In a rehearsal, the director might bang out a part on the piano. Many have posted on the forums that they send music to their choir members that uses woodwinds."
Yes, I use Oboe, Harmonica, F Horn, and Bassoon (Soprano, Alto, Tenor, and Bass, respectively). They sound better than ANY voice sound I've heard in ANY program, while still sounding acceptably like human voice. At the same time, the vocalist doesn't get suckered into listening to the sound and inadvertently picking up intonation from it. It's melody only.
In reply to > "think about using a… by TheHutch
Personally, I don't think WW's sound remotely like a human voice. How does one pick up intonation?
In reply to Personally, I don't think WW… by bobjp
I don't know what "WW" means.
The MS Basic sounds (Is that what they're being called now? I can't remember.) or the MuseHub sounds are close enough to actual human voices that I can see a possibility of singers (especially singers with less experience or less training) inadvertently imitating the sound (the "intonation") of the MuseScore playback. Not invariably a Bad Thing, but probably so.
In reply to I don't know what "WW" means… by TheHutch
WW is a longstanding abbreviation for woodwinds.
MU4 has always differentiated between Basic sounds and Muse sounds. Basic sounds being the sounds that are included with MU4 that are a port over from MU3. Muse sounds are downloaded from the HUB.
I'm not so sure that less experienced singers can pick up on pitch (which is what "intonation" deals more with) problems in Muse sounds. Or any sounds. Sound quality may include being in tune (intonation). But there is much more to sound quality.
Something else to try. Turn on the metronome and listen to one of the parts. I can't hear any late notes, In fact sometimes, 1/4 notes are early.
Same with all the voices together.