I hope this doesn’t come across as hijacking the thread. This is not ment to be a comparison in quality between this mockup and the one from Franck, since these are very different approaches. I just wanted to take the opportunity to share how my template sounds and applied it to the score from Beethoven that was still on my computer.
I’m not aiming to discuss the overall balance of the mockup, but rather its performance.
The setup uses Audio Modeling SWAM instruments (the timpani are from VSL).
First, the downsides:
Even with a lot of work — for example, the first violins consist of three section instruments and two solo violins, the second violins of three sections and one solo, all grouped in Vienna Ensemble Pro — trained ears can still recognize the synthetic nature of the sound.
Now, the advantages, which are the main reason I use only performance-based instruments (SWAM, Sample Modeling, Aaron Venture, etc.) in Dorico:
-
I have an Expression Map that works reliably.
-
I can easily adapt it for each piece: if I want shorter short notes, I can adjust that in the “NoteLength Conditions.” If I want more bite at the start of the note, I simply increase bow pressure.
-
Because there are no dynamic layers, you don’t hear any sample switching during crescendos (the first 30 seconds of Part 2 demonstrate this nicely).
-
The full orchestra takes up only a few hundred megabytes.
-
The earlier disadvantage is also an advantage: I can freely decide whether I want 1, 6, 24, or 32 violins — or anything in between — and with a bit of programming, I can even create convincing divisi passages.
-
Since every articulation originates from the same modeled sound, the result is extremely consistent. Even if you listen to a passage with long notes, fast runs, and short spiccatos in isolation, it still feels like a single coherent performance.
-
By combining Dorico’s playback options, Expression Maps, and the ability to manipulate MIDI data directly in the score, you can achieve very expressive results.
I believe that once Dorico implements proper humanization for long notes, this approach will be almost unbeatable (and the sounds themselves will continue to improve — and if SWAM isn’t to one’s taste, there are alternatives).
What do you think? Have any of you created an Expression Map using samples that works similarly well — and, being sample-based, perhaps sounds even better?