Thanks for sharing your work. It definitely has a lot of melodic and harmonic movement going on, which is great. I think it is indeed your best so far.
On first hearing, I sense a lot of details and structures that probably don't stand out as clearly as you intended. Maybe you experience the same challenges as I do.
I'm working on cinematic/orchestral pieces as examples, on and off as time allows. One thing totally freaks me out every time I do this. The many subtleties in melody and instrumentation that I hear at work are blurred away the next day. Or something else gets into the foreground that I didn't expect. Or worst, everything blends into a porridge that an unbiased listener can't discern anymore.
Showing your work to others, preferably the ones that don't shy away from unfiltered critique (friends, family), always is a confrontation with the realities of perception. I dread it like cold showers in the morning.
Reducing complexity doesn't always bring the desired effect right away. Some things just refuse to stand out no matter what. Especially fast movements, which are supposed to give the whole thing contrast and drama. I figure this carving out of elements into the foreground and assigning instruments different roles that complement each other is a matter of years of experience that I simply don't have. Synfire now puts this within reach, but it still takes time.
It's a challenge with every art, but music seems to be the hardest because it is so subjective.
I wish we had a more suitable social medium where composers can show their work and progress and make this journey a collective experience. Will check some options.
<I'm working on cinematic/orchestral pieces as examples, on and off as time allows. One thing totally freaks me out every time I do this. The many subtleties in melody and instrumentation that I hear at work are blurred away the next day. Or something else gets into the foreground that I didn't expect. Or worst, everything blends into a porridge that an unbiased listener can't discern anymore.>
Foreground, Middleground and Background
(course material example)
Now that we have separated the layers, let’s look at how you can structure these layers into the foreground, middleground, and background. You don’t want all your elements to be the center of attention. Maybe your chords will be the background element, supporting your main melody. Or you have an ostinato that you want to be playing in the background, with other elements being in focus.
(from course) Another common problem I see among a lot of beginning composers is that they add too many elements to their compositions without structuring them properly. The result sounds unfocused, messy and rather chaotic. This is something we talked about in the counter melody chapter, but the same principles apply to any new element you want to add to your track, including harmonies, background textures and so on.
How to check easy the separation of the instruments in Synfire ?
Frequency spectrums, attack and release times, dynamics, harmony relationships, leading and following roles, counter movements, ...>
Are ordinary instruments for classical composing, so the difficulty is not too bad
Totally agree with you that mixing and mastering also play an important role.
I didn't mention it to keep the subject simple.
Separation of instruments is already a good start togehter with the foreground, middle ground and background issue.
I find the mixing and mastering process quite difficult. For this piece I recorded it straight from Synfire, because when I transferred via MIDI to Studio One, volume messages were getting introduced and changing even though not happening in Synfire. I couldn't figure out what was going on, and couldn't seem to adjust this in Studio One.
I like the foreground, midground approach. I think I'll use that in my next track.
I also think that orchestration can help a lot with clarity. But that does require a lot of knowledge which I certainly don't have. Thanks for the comments above.
EDIT: I think a might call my next track, 'Porridge', as a tribute to Andre ;)
I find the mixing and mastering process quite difficult.
I feel the same way. I hate it. But not doing it is not an option. The problem is generally when you make one instrument louder, something else disappears in the background.
In this case, I think it would have been better if the fast string runs at the beginning, which build up the piece well but are barely audible, were much louder. Of course, this would have been at the expense of the children's choir, but the choir could have been pushed to the foreground later. So first introduce the children's choir from a distance (also with more reverb or maybe a delay effect) and then gradually bring it to the foreground. A few bass runs, which bring even more drama to the whole thing, probably wouldn't have hurt either.
Funny you think it's a childrens choir. It's actually the Rhodope 2 Bulgarian ethnic women's choir from Strezov. In this configuration it does indeed sound like Strezov Arva, which is their highly regarded children's choir. I think it's the lack of warbling in the singing style.
On the topic of mixing and mastering, I really think the solution here is to make it possible for those that are particularly skilled at mixing to engage on this. I was hoping to create stems of some of my tracks using a service like Splice, but licensing legalities from sample companies live in the dark ages I believe.
Synfire already allows a degree of remixing at the MIDI level by people sharing libraries. If only this concept could also extend to the production side, imagine what could be achieved collectively.
I really think the solution here is to make it possible for those that are particularly skilled at mixing to engage on this.
That would certainly not be me. I always fiddle endlessly with this mixing stuff and never know how it's right. And then someone comes along and says, this or that is terrible, it can't stay like that. It's a nightmare subject.
This "volume battle" that juergen describes is quite common.
We might learn something from dance producers: If multiple parts don't work together at the same time, chain them after each other in a sequence.
Every part gets a place to stand out while others take a backseat. Music is a sequential medium after all. This also brings contrast and narrative structure. Each "slice" can be as short as a beat or as long as a dozen measures.
> separated the layers, let’s look at how you can structure these layers into the foreground, middleground, and background
This is one of the reasons why I keep asking for having octaves also shown in the output parameter, or -- even better -- some proper pianoroll notation.
Currently, AFAIK there is no way to visually check the pitch registers of different layers of the resulting music, and so there is no way to visually check whether different elements use different pitch registers or whether it is all mashed into overlapping bands. Of course, we can also use our ears, but it can be much quickler to check this visually (e.g., visually I can more quickly jump around in the time line).
Ideally, the pitches shown at the beginning of the output parameter would not "scroll away" but stay as a reference when scrolling through the music. I am not aware of any other DAW where the pitch references at the left-hand side can be "scrolled away", because that just makes reading pitches unnecessarily harder.
BTW: Showing octave numbers in the pitch ranges of instruments would also be helpful.
> It's not that simple. Frequency spectrums, attack and release times, ...
No, its not that simple, but simplifying the pitch separation could be a start. ;-)
> Currently, AFAIK there is no way to visually check the pitch registers of different layers
Update: I found at least a workaround: when clicking on a note in the Output parameter, the parameter inspector at the right hand side displays its note name, including its octave. While not exactly inconvenient, at least it is possible to find the octaves of events in the Output parameter. Thanks!
<No, its not that simple, but simplifying the pitch separation could be a start. ;-)>
This is something I have pointed out before here on forum, but apparently the urgency is not seen.;-)
Another idea was from me to use the Divisimate program in Synfire
The prerequisite for this was that Synfire supports multitrack recording for 32 tracks ?
MIDI pitch does not map to frequency for all instruments alike. Octaves of different instruments may overlap on the keyboard. Putting them side by side in one piano roll does not make sense. If anything, this is only useful within the same instrument, which Output already does.
There is also not enough room to display the whole pitch range of an orchestra. Vertical scrolling inside every track is a total no no. The UI is already fiddly enough. I'm always glad I don't have to scroll because views scale to content automatically. I get that static MIDI pitch may be an exception, but adding a vertical scrollbar to the parameter view will be extremely annoying with the rest of parameters.
The whole point of prototyping is to get rid of this cumbersome micro-management. There must be a better way to indicate range overlap. I have no particular idea yet, though.
Back to topic: Ruchir, I think your new version is much clearer.
<The whole point of prototyping is to get rid of this cumbersome micro-management. There must be a better way to indicate range overlap. I have no particular idea yet, though.>
About time you started paying attention there then :-)
> MIDI pitch does not map to frequency for all instruments alike.
Mostly it does with electronic instruments, and if it does not then users who set up their Synfire projects know the exceptions. BTW: In traditional scores with all sorts of transposing instruments and clefs, the octaves are of course always shown, and score readers then deduce the correct pitches from that.
Sorry, but is this really an argument for not showing octave information? Is there any DAW out there that does not show octave information for such reasons? Sorry for getting a bit edgy here, but I cannot really believe that we have to even argue here for having octave information displayed.
> Vertical scrolling inside every track is a total no no.
The scrolling problem I mentioned occurs with the existing horizonal scrolling: the pitches are shown only at the beginning of the output parameter, and with horizontal scrolling these are "scrolled away" -- unlike all the DAW and notation software I know of that display pitches.
> There must be a better way to indicate range overlap.
That would be welcome.
> The whole point of prototyping is to get rid of this cumbersome micro-management.
Now, this is a big question, that would deserve a separate thread.
I think I fully get your point here: that you want Synfire users to work on a higher level (that plain manual composition). I get that, tor exactly this reason, I have been using and developing algorithmic composition software for decades. However, while I really like to hand low-level musical decisions to a machine, I would like to always retain the right to preserve the aesthetic oversight. E.g., I like to compare slightly different solutions to see what I like better.
Also, remember that various musical aspects that directly affect pitches are not even controlled by Synfire (e.g., the separation of pitch ranges for clarity we just discussed; instrumental limitations -- e.g. that strings can produce double stops only with certain pitch combinations; pitch intervals between important points in music like melodic peaks that are relatively far away etc.). So, even as a Synfire user I have to take care of such issues by hand anyway -- and can still enjoy the tremendous help of software like Synfire.
I therefore always basically proofread and manually edit my algorithmically composed music -- whether created with Synfire or other software. Judging from feedback here in the forum, I am not the only one doing that here, and doing so is indeed a time-honoured venerable tradition (e.g., there is some old literature by G.M. Koenig on this topic).
While I may do some of this manual editing later outside of Synfire, I would prefer doing as much as possible of it directly within Synfire: because then I can do such editing by still working on a higher level by manually adjusting various Synfire settings, instead of doing all low-level decisions fully by hand.
Does the point of such composition proofreading and editing make sense for you?
PS: Apologies to Ruchir for somewhat hijacking his thread here.
It's not about octave information (that information is already there). It's about cramming six or more octaves linearly into a piano roll, which is impossible to oversee without a lot of vertical scrolling. This information can't even be edited.
And yes, this discussion should be held in a separate thread.
From your response I do not really know whether we are really on the same page here. So, I respond once more, but then shut up.
All I am suggesting is that the pitch information in the output parameter can be made better readable, so that it can be used for essentially musically proofreading to inform further editing. Of course, the output parameter cannot be edited directly, but it could be much more useful if (i) at least the octaves would be shown, but (ii) ideally it would be more like a pianoroll notation.
In pianoroll notation, it is (with a little practice) possible to directly read intervals, chord voicings etc. With the current format -- which is just a plain grid where all pitches look the same -- this is next to impossible, even with practice.
Vertical scrolling is currently not supported, and while that might indeed be handy, I am not suggesting adding that, because more octaves can be shown at least when the track sheet is hidden.
For completeness, attached are two screen shots, one showing a Synfire output example, and one with the same data in a piano rool. To be clear, some information is of course already more clearn in Synfire (e.g., the colour-coding of the pitch symbols/roles).
Nevertheless, in the pianoroll notation I can directly and quickly read intervals between arbitrary notes (even without all those note name labels, that is some over-the-top default on my system), but in the Synfire output example I cannot do that really. In my view, the main thing missing is simply different background gray tones to be able to easily distinguish between white and back keys on the keyboard.
Apologies, if this was already all clear anyway. Just wanted to make sure. I should also add that I greatly value your willingness to engage with users in these matters, but that of course any design decisions are only made by the Synfire team.
Comments
Do., 22.12.2022 - 11:48 Permalink
Thanks for sharing your work. It definitely has a lot of melodic and harmonic movement going on, which is great. I think it is indeed your best so far.
On first hearing, I sense a lot of details and structures that probably don't stand out as clearly as you intended. Maybe you experience the same challenges as I do.
I'm working on cinematic/orchestral pieces as examples, on and off as time allows. One thing totally freaks me out every time I do this. The many subtleties in melody and instrumentation that I hear at work are blurred away the next day. Or something else gets into the foreground that I didn't expect. Or worst, everything blends into a porridge that an unbiased listener can't discern anymore.
Showing your work to others, preferably the ones that don't shy away from unfiltered critique (friends, family), always is a confrontation with the realities of perception. I dread it like cold showers in the morning.
Reducing complexity doesn't always bring the desired effect right away. Some things just refuse to stand out no matter what. Especially fast movements, which are supposed to give the whole thing contrast and drama. I figure this carving out of elements into the foreground and assigning instruments different roles that complement each other is a matter of years of experience that I simply don't have. Synfire now puts this within reach, but it still takes time.
It's a challenge with every art, but music seems to be the hardest because it is so subjective.
I wish we had a more suitable social medium where composers can show their work and progress and make this journey a collective experience. Will check some options.
Do., 22.12.2022 - 13:52 Permalink
<I'm working on cinematic/orchestral pieces as examples, on and off as time allows. One thing totally freaks me out every time I do this. The many subtleties in melody and instrumentation that I hear at work are blurred away the next day. Or something else gets into the foreground that I didn't expect. Or worst, everything blends into a porridge that an unbiased listener can't discern anymore.>
Foreground, Middleground and Background
(course material example)
Now that we have separated the layers, let’s look at how you can structure these layers into the foreground, middleground, and background. You don’t want all your elements to be the center of attention. Maybe your chords will be the background element, supporting your main melody. Or you have an ostinato that you want to be playing in the background, with other elements being in focus.
(from course) Another common problem I see among a lot of beginning composers is that they add too many elements to their compositions without structuring them properly. The result sounds unfocused, messy and rather chaotic. This is something we talked about in the counter melody chapter, but the same principles apply to any new element you want to add to your track, including harmonies, background textures and so on.
How to check easy the separation of the instruments in Synfire ?
Do., 22.12.2022 - 14:25 Permalink
It's not that simple.
Frequency spectrums, attack and release times, dynamics, harmony relationships, leading and following roles, counter movements, ...
This thread is probably not the place to discuss this in depth. I just wanted to touch the topic as it came to mind.
Do., 22.12.2022 - 14:33 Permalink
And mixing. Never underestimate the importance of mixing and mastering.
Do., 22.12.2022 - 15:07 Permalink
<It's not that simple.
Frequency spectrums, attack and release times, dynamics, harmony relationships, leading and following roles, counter movements, ...>
Are ordinary instruments for classical composing, so the difficulty is not too bad
Totally agree with you that mixing and mastering also play an important role.
I didn't mention it to keep the subject simple.
Separation of instruments is already a good start togehter with the foreground, middle ground and background issue.
Do., 22.12.2022 - 17:01 Permalink
I find the mixing and mastering process quite difficult. For this piece I recorded it straight from Synfire, because when I transferred via MIDI to Studio One, volume messages were getting introduced and changing even though not happening in Synfire. I couldn't figure out what was going on, and couldn't seem to adjust this in Studio One.
I like the foreground, midground approach. I think I'll use that in my next track.
I also think that orchestration can help a lot with clarity. But that does require a lot of knowledge which I certainly don't have. Thanks for the comments above.
EDIT: I think a might call my next track, 'Porridge', as a tribute to Andre ;)
Do., 22.12.2022 - 19:22 Permalink
I feel the same way. I hate it. But not doing it is not an option. The problem is generally when you make one instrument louder, something else disappears in the background.
In this case, I think it would have been better if the fast string runs at the beginning, which build up the piece well but are barely audible, were much louder. Of course, this would have been at the expense of the children's choir, but the choir could have been pushed to the foreground later. So first introduce the children's choir from a distance (also with more reverb or maybe a delay effect) and then gradually bring it to the foreground. A few bass runs, which bring even more drama to the whole thing, probably wouldn't have hurt either.
Overall, however, it is of course a great track.
Do., 22.12.2022 - 19:48 Permalink
Funny you think it's a childrens choir. It's actually the Rhodope 2 Bulgarian ethnic women's choir from Strezov. In this configuration it does indeed sound like Strezov Arva, which is their highly regarded children's choir. I think it's the lack of warbling in the singing style.
On the topic of mixing and mastering, I really think the solution here is to make it possible for those that are particularly skilled at mixing to engage on this. I was hoping to create stems of some of my tracks using a service like Splice, but licensing legalities from sample companies live in the dark ages I believe.
Synfire already allows a degree of remixing at the MIDI level by people sharing libraries. If only this concept could also extend to the production side, imagine what could be achieved collectively.
Do., 22.12.2022 - 20:18 Permalink
That would certainly not be me. I always fiddle endlessly with this mixing stuff and never know how it's right. And then someone comes along and says, this or that is terrible, it can't stay like that. It's a nightmare subject.
Do., 22.12.2022 - 22:05 Permalink
This "volume battle" that juergen describes is quite common.
We might learn something from dance producers: If multiple parts don't work together at the same time, chain them after each other in a sequence.
Every part gets a place to stand out while others take a backseat. Music is a sequential medium after all. This also brings contrast and narrative structure. Each "slice" can be as short as a beat or as long as a dozen measures.
Do., 22.12.2022 - 22:37 Permalink
Based on the comments, I've added a bit of 'less is more' to this track and created a new version.
Hoping this is less muddy.
EDIT: I'll reupload to the original video on Youtube and replace the link.
Fr., 23.12.2022 - 12:52 Permalink
> separated the layers, let’s look at how you can structure these layers into the foreground, middleground, and background
This is one of the reasons why I keep asking for having octaves also shown in the output parameter, or -- even better -- some proper pianoroll notation.
Currently, AFAIK there is no way to visually check the pitch registers of different layers of the resulting music, and so there is no way to visually check whether different elements use different pitch registers or whether it is all mashed into overlapping bands. Of course, we can also use our ears, but it can be much quickler to check this visually (e.g., visually I can more quickly jump around in the time line).
Ideally, the pitches shown at the beginning of the output parameter would not "scroll away" but stay as a reference when scrolling through the music. I am not aware of any other DAW where the pitch references at the left-hand side can be "scrolled away", because that just makes reading pitches unnecessarily harder.
BTW: Showing octave numbers in the pitch ranges of instruments would also be helpful.
> It's not that simple. Frequency spectrums, attack and release times, ...
No, its not that simple, but simplifying the pitch separation could be a start. ;-)
Fr., 23.12.2022 - 13:02 Permalink
> (from course)
Would you mind sharing what course material you are referring to here? Are you teaching composition? Thanks!
Fr., 23.12.2022 - 13:23 Permalink
> Currently, AFAIK there is no way to visually check the pitch registers of different layers
Update: I found at least a workaround: when clicking on a note in the Output parameter, the parameter inspector at the right hand side displays its note name, including its octave. While not exactly inconvenient, at least it is possible to find the octaves of events in the Output parameter. Thanks!
Note to future self: middle C is C3 in Synfire.
Fr., 23.12.2022 - 13:49 Permalink
<No, its not that simple, but simplifying the pitch separation could be a start. ;-)>
This is something I have pointed out before here on forum, but apparently the urgency is not seen.;-)
Fr., 23.12.2022 - 13:51 Permalink
Its from Evenant : Cinematic Music: From Idea To Finished Recording - Evenant
Fr., 23.12.2022 - 14:04 Permalink
Another idea was from me to use the Divisimate program in Synfire
The prerequisite for this was that Synfire supports multitrack recording for 32 tracks ?
Divisimate - Realtime Virtual Orchestration
Fr., 23.12.2022 - 16:12 Permalink
MIDI pitch does not map to frequency for all instruments alike. Octaves of different instruments may overlap on the keyboard. Putting them side by side in one piano roll does not make sense. If anything, this is only useful within the same instrument, which Output already does.
There is also not enough room to display the whole pitch range of an orchestra. Vertical scrolling inside every track is a total no no. The UI is already fiddly enough. I'm always glad I don't have to scroll because views scale to content automatically. I get that static MIDI pitch may be an exception, but adding a vertical scrollbar to the parameter view will be extremely annoying with the rest of parameters.
The whole point of prototyping is to get rid of this cumbersome micro-management. There must be a better way to indicate range overlap. I have no particular idea yet, though.
Back to topic: Ruchir, I think your new version is much clearer.
Fr., 23.12.2022 - 16:23 Permalink
<The whole point of prototyping is to get rid of this cumbersome micro-management. There must be a better way to indicate range overlap. I have no particular idea yet, though.>
About time you started paying attention there then :-)
Fr., 23.12.2022 - 17:13 Permalink
> MIDI pitch does not map to frequency for all instruments alike.
Mostly it does with electronic instruments, and if it does not then users who set up their Synfire projects know the exceptions. BTW: In traditional scores with all sorts of transposing instruments and clefs, the octaves are of course always shown, and score readers then deduce the correct pitches from that.
Sorry, but is this really an argument for not showing octave information? Is there any DAW out there that does not show octave information for such reasons? Sorry for getting a bit edgy here, but I cannot really believe that we have to even argue here for having octave information displayed.
> Vertical scrolling inside every track is a total no no.
The scrolling problem I mentioned occurs with the existing horizonal scrolling: the pitches are shown only at the beginning of the output parameter, and with horizontal scrolling these are "scrolled away" -- unlike all the DAW and notation software I know of that display pitches.
> There must be a better way to indicate range overlap.
That would be welcome.
> The whole point of prototyping is to get rid of this cumbersome micro-management.
Now, this is a big question, that would deserve a separate thread.
I think I fully get your point here: that you want Synfire users to work on a higher level (that plain manual composition). I get that, tor exactly this reason, I have been using and developing algorithmic composition software for decades. However, while I really like to hand low-level musical decisions to a machine, I would like to always retain the right to preserve the aesthetic oversight. E.g., I like to compare slightly different solutions to see what I like better.
Also, remember that various musical aspects that directly affect pitches are not even controlled by Synfire (e.g., the separation of pitch ranges for clarity we just discussed; instrumental limitations -- e.g. that strings can produce double stops only with certain pitch combinations; pitch intervals between important points in music like melodic peaks that are relatively far away etc.). So, even as a Synfire user I have to take care of such issues by hand anyway -- and can still enjoy the tremendous help of software like Synfire.
I therefore always basically proofread and manually edit my algorithmically composed music -- whether created with Synfire or other software. Judging from feedback here in the forum, I am not the only one doing that here, and doing so is indeed a time-honoured venerable tradition (e.g., there is some old literature by G.M. Koenig on this topic).
While I may do some of this manual editing later outside of Synfire, I would prefer doing as much as possible of it directly within Synfire: because then I can do such editing by still working on a higher level by manually adjusting various Synfire settings, instead of doing all low-level decisions fully by hand.
Does the point of such composition proofreading and editing make sense for you?
PS: Apologies to Ruchir for somewhat hijacking his thread here.
Fr., 23.12.2022 - 17:19 Permalink
> Its from Evenant
Thanks for sharing.
Fr., 23.12.2022 - 20:16 Permalink
It's not about octave information (that information is already there). It's about cramming six or more octaves linearly into a piano roll, which is impossible to oversee without a lot of vertical scrolling. This information can't even be edited.
And yes, this discussion should be held in a separate thread.
Fr., 23.12.2022 - 22:19 Permalink
From your response I do not really know whether we are really on the same page here. So, I respond once more, but then shut up.
All I am suggesting is that the pitch information in the output parameter can be made better readable, so that it can be used for essentially musically proofreading to inform further editing. Of course, the output parameter cannot be edited directly, but it could be much more useful if (i) at least the octaves would be shown, but (ii) ideally it would be more like a pianoroll notation.
In pianoroll notation, it is (with a little practice) possible to directly read intervals, chord voicings etc. With the current format -- which is just a plain grid where all pitches look the same -- this is next to impossible, even with practice.
Vertical scrolling is currently not supported, and while that might indeed be handy, I am not suggesting adding that, because more octaves can be shown at least when the track sheet is hidden.
For completeness, attached are two screen shots, one showing a Synfire output example, and one with the same data in a piano rool. To be clear, some information is of course already more clearn in Synfire (e.g., the colour-coding of the pitch symbols/roles).
Nevertheless, in the pianoroll notation I can directly and quickly read intervals between arbitrary notes (even without all those note name labels, that is some over-the-top default on my system), but in the Synfire output example I cannot do that really. In my view, the main thing missing is simply different background gray tones to be able to easily distinguish between white and back keys on the keyboard.
Apologies, if this was already all clear anyway. Just wanted to make sure. I should also add that I greatly value your willingness to engage with users in these matters, but that of course any design decisions are only made by the Synfire team.
Fr., 23.12.2022 - 22:21 Permalink
Sorry, here are the screen shots.