Hi,
Very impressive work, I haven't tried the template but I'm impressed with the distinct timbres of the various sounds! Thanks for all your efforts on this!
Orchestra Template Test
Moderators: MattKingUSA, khz
- Michael Willis
- Established Member
- Posts: 1458
- Joined: Mon Oct 03, 2016 3:27 pm
- Location: Rocky Mountains, North America
- Has thanked: 70 times
- Been thanked: 167 times
- Contact:
Re: Orchestra Template Test
Hmm, I experienced this problem before and forgot about it! I tried to ask about it on the Ardour forum, but didn't get any answers. Last time I found a way to work around the problem, but I don't think it will work on this one.Rainmak3r wrote:As a side note, one thing I forgot to mention last time is that your template has hardcoded paths to a local folder on your machine for the "template-dir" property, and that applies to all tracks: I thought this was normal and harmless (I see the same in another template I have installed), but if I don't manually fix the template-dir to reflect the real path on my laptop, none of the SFZ are loaded. I guess it's normal for Ardour to save your path when saving the template, but maybe there should be a placeholder strng on the version you have on github, and instructions in the README on how to replace it? (e.g., with an inline sed). There may be a variable you can use in the template to refer to the local template folder no matter the installation, but I'm not familiar with the syntax unfortunately.
https://discourse.ardour.org/t/template ... -dir/89283
Again, for being a fairly quick import from MuseScore, this sounds good to me!Rainmak3r wrote:Just exported a new version of the waltz using the v2-tweaks template: https://drive.google.com/open?id=1jp42N ... 48A9Jraza-
milo wrote:That sounds really good, Michael! I don't have my nice monitor speakers or headphones here in my office, but on my little pc speakers it sounds pretty good. I'll give it a listen on the nicer system when I get home.
Thanks milo and GMaq! To be honest I'm still not totally satisfied with the sound. I had some brass players listen to it yesterday, and they told me that the brass part isn't convincing. However I think there probably isn't much more I can do without getting a more professional set of orchestral samples.GMaq wrote:Very impressive work, I haven't tried the template but I'm impressed with the distinct timbres of the various sounds! Thanks for all your efforts on this!
- Rainmak3r
- Established Member
- Posts: 892
- Joined: Sat Mar 02, 2019 12:24 pm
- Has thanked: 49 times
- Been thanked: 180 times
- Contact:
Re: Orchestra Template Test
Hi Michael,
not sure if I should be asking this here, another post, or another place entirely, but I thought I'd add this in response to this discussion as it's motivated by your efforts on the template.
I've been working on a few other, more complex, symphonic pieces, and when I tried to follow the same approach I used for the waltz (export to MIDI from MuseScore, import in Ardour), the results were much worse this time. I did a bit of research on both MuseScore and VPO, and I think I now have a slightly better understanding of the different SFZ options that are available.
Apparently, the -PERF scripts won't work as expected with MuseScore, which uses velocity to control volume (unlike the -PERF scripts that use the mod wheel). When I tried the mod-wheel versions, results were a bit better, but of course with broken volumes (which you did warn me about, saying mixing levels should have been adjusted as they're probably tailored for the -PERF ones). In the process, I finally understood how the -KS versions work, and they do seem more flexible indeed, as they allow for a dynamic "swap" of samples when you use a specific (an unused for the instrument) key: this seems perfect for tracks generated by MuseScore, as it allows me not to have to duplicate tracks for pizzicato and tremolo parts (as I had to do with the waltz), but just keep on using the existing one; I just need to insert a key switch where I know I need one, and it should do the job for me. As such, in the next few weeks I plan to adapt your template to use those scripts, and test whether there is indeed an improvement.
That said, I suspect some of the -KS scripts are broken, or maybe there's something else I don't understand, which is what I wanted to ask you about. When I manually load them via sfizz_jack to test them with my keyboard, while viola-SEC-KS-C2.sfz works fine, for instance, 1st-violin-SEC-KS-C2.sfz and 2nd-violin-SEC-KS-C2.sfz have an incredibly low volume. I had a look at the SFZ format to try and understand the syntax, but apparently the volume properties for the samples seem the same as the non-KS versions (even though I didn't really check them all, which I probably should). Considering I found a few exchanges between you and the author, where you apparently did actually test them all (or most of them) to contribute fixes, are you aware of this issue? I haven't tested them all, so I don't know if the violins section are the only ones affected, but it would be helpful to know if there was indeed a regression from the previous version (I'm using 3.2 right now), and what to look for in case.
Thanks!
not sure if I should be asking this here, another post, or another place entirely, but I thought I'd add this in response to this discussion as it's motivated by your efforts on the template.
I've been working on a few other, more complex, symphonic pieces, and when I tried to follow the same approach I used for the waltz (export to MIDI from MuseScore, import in Ardour), the results were much worse this time. I did a bit of research on both MuseScore and VPO, and I think I now have a slightly better understanding of the different SFZ options that are available.
Apparently, the -PERF scripts won't work as expected with MuseScore, which uses velocity to control volume (unlike the -PERF scripts that use the mod wheel). When I tried the mod-wheel versions, results were a bit better, but of course with broken volumes (which you did warn me about, saying mixing levels should have been adjusted as they're probably tailored for the -PERF ones). In the process, I finally understood how the -KS versions work, and they do seem more flexible indeed, as they allow for a dynamic "swap" of samples when you use a specific (an unused for the instrument) key: this seems perfect for tracks generated by MuseScore, as it allows me not to have to duplicate tracks for pizzicato and tremolo parts (as I had to do with the waltz), but just keep on using the existing one; I just need to insert a key switch where I know I need one, and it should do the job for me. As such, in the next few weeks I plan to adapt your template to use those scripts, and test whether there is indeed an improvement.
That said, I suspect some of the -KS scripts are broken, or maybe there's something else I don't understand, which is what I wanted to ask you about. When I manually load them via sfizz_jack to test them with my keyboard, while viola-SEC-KS-C2.sfz works fine, for instance, 1st-violin-SEC-KS-C2.sfz and 2nd-violin-SEC-KS-C2.sfz have an incredibly low volume. I had a look at the SFZ format to try and understand the syntax, but apparently the volume properties for the samples seem the same as the non-KS versions (even though I didn't really check them all, which I probably should). Considering I found a few exchanges between you and the author, where you apparently did actually test them all (or most of them) to contribute fixes, are you aware of this issue? I haven't tested them all, so I don't know if the violins section are the only ones affected, but it would be helpful to know if there was indeed a regression from the previous version (I'm using 3.2 right now), and what to look for in case.
Thanks!
- Michael Willis
- Established Member
- Posts: 1458
- Joined: Mon Oct 03, 2016 3:27 pm
- Location: Rocky Mountains, North America
- Has thanked: 70 times
- Been thanked: 167 times
- Contact:
Re: Orchestra Template Test
That's quite possible. I never tested the KS scripts, as I prefer separate tracks per articulation. The sfizz developers are fairly responsive; try reporting this in their discord channel, or as an issue on GitHub. Be prepared to send an example Ardour session.Rainmak3r wrote:That said, I suspect some of the -KS scripts are broken
- Rainmak3r
- Established Member
- Posts: 892
- Joined: Sat Mar 02, 2019 12:24 pm
- Has thanked: 49 times
- Been thanked: 180 times
- Contact:
Re: Orchestra Template Test
Oh, right, I didn't think it might be an issue with sfizz... I'll give it a try with qsampler too, and if it works there I'll contact the sfizz guys. Thanks for the quick answer! (and apologies for misspelling your last name from time to time, I just noticed it's Willis and NOT Wellis as I thought )Michael Willis wrote:That's quite possible. I never tested the KS scripts, as I prefer separate tracks per articulation. The sfizz developers are fairly responsive; try reporting this in their discord channel, or as an issue on GitHub. Be prepared to send an example Ardour session.Rainmak3r wrote:That said, I suspect some of the -KS scripts are broken
- Michael Willis
- Established Member
- Posts: 1458
- Joined: Mon Oct 03, 2016 3:27 pm
- Location: Rocky Mountains, North America
- Has thanked: 70 times
- Been thanked: 167 times
- Contact:
Re: Orchestra Template Test
From what I've been able to determine, it's an old English name that refers to somebody who owned a well and/or was responsible for the maintenance thereof, so Wellis isn't too inaccurate.Rainmak3r wrote:Willis and NOT Wellis
- Rainmak3r
- Established Member
- Posts: 892
- Joined: Sat Mar 02, 2019 12:24 pm
- Has thanked: 49 times
- Been thanked: 180 times
- Contact:
Re: Orchestra Template Test
Mh, I just tried again and it worked as expected now... probably a temporary issue in my setup yesterday, I guess. Should it happen again, I'll make sure to make more tests with LinuxSampler as well.Rainmak3r wrote:Oh, right, I didn't think it might be an issue with sfizz... I'll give it a try with qsampler too, and if it works there I'll contact the sfizz guys. Thanks for the quick answer! (and apologies for misspelling your last name from time to time, I just noticed it's Willis and NOT Wellis as I thought )
- psyocean
- Established Member
- Posts: 651
- Joined: Thu Oct 13, 2016 4:54 pm
- Location: Saint-Petersburg, Russia
- Has thanked: 2 times
- Been thanked: 89 times
- Contact:
Re: Orchestra Template Test
It looks like a great project! Sorry, I'm not good in English, as I understand this is a SFZ play-edit tool. Necessary thing! After the repositories with Linuxsampler was died, I now stay without SFZ libraries. Perhaps this can revive them - the glory and power of SFZ tools! If it will be placed in some deb-repository - it be so useful
Guitar and synth tales... https://www.youtube.com/user/Psyocean/
- Rainmak3r
- Established Member
- Posts: 892
- Joined: Sat Mar 02, 2019 12:24 pm
- Has thanked: 49 times
- Been thanked: 180 times
- Contact:
Re: Orchestra Template Test
FYI, I've started working on that CC support in both MuseScore and VPO (and in the future, hopefully, Ardour):Rainmak3r wrote: ↑Mon Feb 24, 2020 1:13 pmMh, I just tried again and it worked as expected now... probably a temporary issue in my setup yesterday, I guess. Should it happen again, I'll make sure to make more tests with LinuxSampler as well.Rainmak3r wrote:Oh, right, I didn't think it might be an issue with sfizz... I'll give it a try with qsampler too, and if it works there I'll contact the sfizz guys. Thanks for the quick answer! (and apologies for misspelling your last name from time to time, I just noticed it's Willis and NOT Wellis as I thought )
https://github.com/lminiero/musescore-vpo-midi-actions
I'll open a separate post in the Recorders/Sequencers section; it's far from complete and has a few issues/limitations, but I'd love to hear your feedback!