Digital Devil Saga 2 - Brahman

[GrapplingHook]

Well then. Here's a thing what I did:

http://tindeck.com/listen/gahg

And one of the first youtube results for the source:

The first night I was working on it, I got super excited and started to think it was really, really good. Then, as time passed, I started to like it less and less. So let's see whether I've just heard it too many times or it's legitimately bad (or good) [or whatever].

[timaeus222]

The strings are missing some humanization, and more reverb would help. Their phrasings aren't smooth, so that breaks their melodic flow. Some automation on the volumes to add swells on sustains would really help the realism. I'm unsure how off the rhythm could be in the intro since there's no rhythmic reference.

The piano sounds rather mechanical, especially near the moment the slider reaches the left of the T in "Track". Each note sounds like it's literally starting over. Its sample quality is kind of low, too. Also, it would really help to adjust the velocities to how a real pianist would play it, overlap notes a bit, and offset the timings a little bit for more realism.

Since the texture of this track is pretty much just piano+strings, it's that much more important to fix those issues. Good start.

This could be a good read.

[GrapplingHook]

Hello! And thanks! I'm still pretty new to this whole music thing, what with the not playing any instruments or having any proper education or even getting started long enough ago to have experience, so tutorial links are more than welcome. Unfortunately for that one, the example mp3s are no longer found where the links point.

When you talk about the piano, there are two different things you could mean, and I don't want to assume I know which one you're referencing since I don't have too much of an ear yet and can't trust that I am or am not hearing right.

Do you mean the lower, harsh bass piano or the higher, softer one? To my ear, it's the lower one that's less human.

I have already done some velocity work (but pretty much no fine timing adjustment) on the soft piano and the strings. The other piano velocities could be tweaked, and timing could certainly use some work all around, but if it's the higher piano that's not doing it then I could go back and push the velocities around more.

The intro is probably going to wind up scrapped, arrangement-wise. I'm (reasonably) happy with the basic structure of the repeating bits (refrain? chorus? some sort of music word...), the piano interlude, the build back into repeating, and the ending, but the intro and the transition to solo piano are... suboptimal.

So you're right to question the intro as-is. I had to muck around a lot to get the coarse timing I wanted instead of leaving it be and I'm still not entirely pleased with it.

And back to the two pianos, which one (if not both) do you think are low quality samples? If the lower one, I can try to fix that up. The base sample doesn't sound quite like that - I just moved a lot of knobs around to give it the "hard" edge I wanted, and it's probably doable without a loss of quality.

Again, thanks.

[timaeus222]

When you talk about the piano, there are two different things you could mean, and I don't want to assume I know which one you're referencing since I don't have too much of an ear yet and can't trust that I am or am not hearing right.

Do you mean the lower, harsh bass piano or the higher, softer one? To my ear, it's the lower one that's less human.

I thought they were the same piano. In that case, yeah, the lower one is less human.

I have already done some velocity work (but pretty much no fine timing adjustment) on the soft piano and the strings. The other piano velocities could be tweaked, and timing could certainly use some work all around, but if it's the higher piano that's not doing it then I could go back and push the velocities around more.

When the time slider reaches in between the T and r of "Track", the Ti-Do-Re-Mi (or Mi-Fa-So-La) notes sounded mechanical. Might as well check both, then!

And back to the two pianos, which one (if not both) do you think are low quality samples? If the lower one, I can try to fix that up. The base sample doesn't sound quite like that - I just moved a lot of knobs around to give it the "hard" edge I wanted, and it's probably doable without a loss of quality.

Again, thanks.

The lower one sounds more low-quality than the other. By quality, I mean the results of the depth of the sampling in however you're using it (a General MIDI soundfont?), but not actually what I simply like better. To conserve space, many soundfonts sample at the C of every octave, and the more realistic ones sample at those C's plus each G, at least. Each pitch recorded is then shifted up and down the keyboard to give all the note timbres, and as you may guess, drastic pitch shifts (C<->G or larger) can warp the tone a lot, especially on the lowest and highest notes on a piano.

So, I'm not saying that a real piano can't sound better than a soundfont, but if the soundfont sampled a piano that's supposed to sound higher quality in timbre than the real piano that was compared to it (i.e. Steinway & Sons vs. Yamaha S Series), then I'd say the soundfont sounds higher quality, but the real piano, of course, sounds more realistic. I'd opt for realism more so than quality any day, but if you can humanize the soundfont, then you've accomplished both parts.

Edited February 12, 2014 by timaeus222

[arglactable]

Cool. Shin Megami Tensei.

Humanization is definitely your biggest problem, pretty much all around.

The piano in the bass is pretty iffy throughout, not just in terms of velocity and timbre changes, but arrangement. I would not recommend using it as a simple bass instrument the way you did (following the melody's rhythm on the root). Adding some movement to that part would go along way towards making it sound not only a bit more believable, but just better. You could just start with the root and the fifth of each chord and expand it from there. I'm also curious what exactly you're using for the sound here, because I know of a couple of free sampled/modeled pianos that would probably be an improvement.

As for the strings, the staccato suffers from the kind of "chugging" effect that's generally caused by utilizing the same staccato string sample in series. It's pretty common with free samples/soundfonts that don't have any sort of round robin support (which is pretty much all of them, especially with strings). Changing up the velocity can help, but getting the variety of timbre from a real performance could prove challenging. I've had some success with LFOs.

I'm not sure exactly what sounds off about the sustained strings, though I would add some harmony in there, as opposed to the unison octaves.

I would also recommend some tasteful reverb, not just to give it a sense of space, but because that added distance from the listener will make some of the more mechanical sounding parts less noticeable in general.

[timaeus222]

I'm not sure exactly what sounds off about the sustained strings, though I would add some harmony in there, as opposed to the unison octaves.

I would also recommend some tasteful reverb, not just to give it a sense of space, but because that added distance from the listener will make some of the more mechanical sounding parts less noticeable in general.

The sustains sound flat. They need some automation to "swell" the volumes and emulate a human's tendency to constantly vary the volume at which they play. Harmonies help the depth of the sound, but they also add the need to vary the note start timings. The more harmonies you add, the more stiffness can be an issue.

Reverb helps, but using only reverb in hopes of "hiding" the mechanical parts is like covering up the problem. It's like saying you solved a math problem in the wrong way and got the right answer, then assumed that getting the right answer means you did the work correctly, when instead you may have made a mistake and one mistake covered up the other.

Edited February 12, 2014 by timaeus222

[arglactable]

I wasn't suggesting using only reverb to hide synthetic-sounding articulation, but it would certainly help an already improved performance sound better. There is only so much you can do with automation to make free strings sound better, especially dry. Fluctuations in velocity and pitch are doable, but fluctuations in timbre that aren't already present in the synth/samples are less so.

[timaeus222]

I wasn't suggesting using only reverb to hide synthetic-sounding articulation, but it would certainly help an already improved performance sound better. There is only so much you can do with automation to make free strings sound better, especially dry. Fluctuations in velocity and pitch are doable, but fluctuations in timbre that aren't already present in the synth/samples are less so.

Gotcha; At the same time though, that post was meant as a piece of info for everyone.