I have searched the entire forum, and yes, so many people are having the same problem, but there has been no solutions.
PROBLEM
In Synthesizer V editor window, when you make a change to one note, the entire track is being live rendered from the beginning to the end, and the software is frozen during the rendering time, you have to wait before you can playback.
Why Synthesizer V is so dumb? Why can’t it simply live render the changed part?
This basically makes the software unusable. Every tiny change you make, you have to wait 10 seconds before you can listen to the change.
I am using an Intel i7 quadcore CPU, which is considered a modern powerful processor.
WORKAROUND
Break a song into 50 tracks, each track only contains one sentence, and it’s so dumb to do it this way.
Use a standard voicebank to do the editing, then switch back to an AI voicebank when everything is done. It’s a dumb method.
I’ve never seen this issue. Normally, the only time the entire track will be re-rendered is if you change something that affects the entire track (like a base parameter value in the Voice panel).
If you open the render panel and modify just one note, you should see only that part of the track be re-rendered:
Please take a look at this project file. It’s very small, only contain 2 sentences. If you change a tiny thing on one note, such as the note’s length, position, lyrics, etc, the whole track is being live rendered, and the green progress bar will drop to 0%, and then slowly goes up to 100%.
if a select a voice (no voice was selected when i opened it) and then see the note blobs. ok does 100%. then changing a note, i see the scan and the change (new blob and gap in the processing display) but i can play it immediately. even on longer song this is the typical wya it processes for me. make a change, re-play it. very seldom do i get delays or skips.
one thing to check - are you using any cloud service for your file system? the blobs are stored on files and if you’re using a cloud service there may be contention and delays as a result.
In my situation, if you try to play it, it will hiccup/stutter. A region in the editor can not be smoothly playbacked, until it’s covered by the green progress bar in the “Render” panel.
Hi there,
I’m here on the macbook and can’t confirm the problem. According to my tests here, SY-V only renders a part around the edited area. no matter whether stand-alone or plugin. No complete rendering noticed. SY-V version 1.11.0.
Hi,
I just tried your file. Very strange what is happening.
As I mentioned below, I cannot confirm this with my projects. I would send the file to support.
Your file does exactly what you describe. As mentioned above, my projects don’t behave like this. only a certain area around the edited one is changed.
I tried your file. It doesn’t happen to me.
It changes very smooth and fast.
But I had it before the switch from the apple intel i7 to my new apple silicon M1.
I’m not at a computer to verify this, but based on the video it looks like it’s rendering everything starting from the nearest silence before your changed note.
This is expected. Unless there are moments of silence, it is one continuous waveform. Cutting it and only re-rendering part of it would risk causing clicks or pops.
You could try increasing the gap between one of the sustained notes and the following breath.
Put differently, I suspect SynthV is avoiding situations where it would otherwise need to crossfade between an old and new render.
If performance is a concern, try adjusting the number of rendering threads in the settings panel. Given that you have a quad-core processor, higher isn’t necessarily better. Perhaps try 4 or 8?
I tried with your project and I see also the rendering effect with all sound processing.
But with on my PC, it is very fast (multi-threaded processing).
So I was not bothered by this behavior.
And I tried another thing with group of notes to isolate (reduce) number of notes to re-render.
The result is quitly different. Only the group of notes is updated. It’s even faster!
See the video on my computer: SynthV-Project1-jfa-r.zip (1,5 Mo)
I tested the group method before, and it worked, only the group region is re-rendered. But it’s quite annoying to created 50 groups for a whole song, sentence by sentence.
As Claire said, only the note before and the note after the changed one should be re-rendered, not the whole track.
My example was really short, only one verse, that’s why you are ok with it.
I assume, if you tested on a full song, doesn’t matter how fast your computer is, you will hiccup on playback.
I tried both 4 threads and 8 threads, with maximum performance. Takes about 6 to 10 seconds delay.
The problem is, if it’s a full song length, nobody can handle it, doesn’t matter how strong the cpu is, you are not supposed to render a full song based on single note change, am I right?
May be your are right about the complete song with huge re-render.
That’s the main reason that my DAW crash sometimes because of huge project. Too many threads.
Especially when you try another vocal bank…
But on my PC, I often use 3/5 tracks on the SynthV project into my DAW.
Groups are used in my case to capture/duplicate the intro/chorus/verse, etc.
Of course, with the displayed running time of your video. I won’t be able to work on my projects.
The silence gap between adjacent notes are more than one quarter long. So, it has nothing to do with trying to avoid crossfades between phrase blocks.
Please see attached video. Pay close attention to the green progress bar inside the Render panel.
You change a single note’s lyrics, then for the next 1 second, things seems to be right, only a gap show in the progress bar, representing where it should be re-rendered.
After 1 second, things went wrong. The full track is being re-rendered.
The funny thing is, the track is only being re-rendered from the changed note to the end of the track.
Anything located before that changed note, will not be re-rendered, so works normally.
In other words, if you changed a note in an early position, then it takes longer to reach to the end, therefore more delay. If you change a note in a later position, then the re-renders take less time to complete with less delay.
I still strongly believe, this is more of a setting/configuration problem.
So many people in this forum have experienced this issue.
It would be a boss move, if you could create a brand new file, import a full-song-length (5 minutes) MIDI to SynthV, the default “la” is enough, then start changing lyrics one note at a time, see if the same re-render problem occurs.
The weird thing is, lots of people are saying: some files are good, some files are problematic.
So, even for the same computer, this problem may happen, and may not happen, depending on the files.
So, it’s nothing to do with the CPU. Even the most powerful CPU would still hiccup on a full-length-song.
