Why do Sequential/DSI synths sound "harsh" to me?

[jdt9517]

Exactly... and I know that back then, DSI changed this range... and I bet it was because of the opposite... people back then wanted it to be brighter....

I was one of those "back then" who wanted the P-5 to be brighter.  Hence why I prefer the P-08 over the P-5 sound wise.

[MoogieBoogieMan]

I figured out why I think it sounds "harsh" to me.  It's not about brightness, I like bright sounds.  I don't hear the "harshness" when I hear a saw wave from Serum when the filter is bypassed.   But I can hear a similar "harshness" when I have the filter enabled and all the way open in Serum.  The filter is still cutting out a bit of the highs even when it's all the way open.  So to my ears, I think the "harshness" that I'm hearing is the Curtis filter affecting the highs.  SO now I'm totally fine with the "harshness" that I was hearing now that I know where it was coming from.  It would be great to be able to bypass the filter, but that doesn't really make a lot of sense on an analog subtractive synth.

[guyaguy]

I figured out why I think it sounds "harsh" to me.  It's not about brightness, I like bright sounds.  I don't hear the "harshness" when I hear a saw wave from Serum when the filter is bypassed.   But I can hear a similar "harshness" when I have the filter enabled and all the way open in Serum.  The filter is still cutting out a bit of the highs even when it's all the way open.  So to my ears, I think the "harshness" that I'm hearing is the Curtis filter affecting the highs.  SO now I'm totally fine with the "harshness" that I was hearing now that I know where it was coming from.  It would be great to be able to bypass the filter, but that doesn't really make a lot of sense on an analog subtractive synth.

I've never heard treble attenuation described as harshness.

[chapterhaus]

I too think that my DSI is a little crisper than other synths; so I did a little test. Backstory: I've been trying to model some Juno 106 patches (borrowing from a friend) on my Mopho X4, and for the most part can get them dialed in. I began with debatably the easiest tone, Patch A11, Brass. A simple saw wave with some chorus. I replicated the patch sans chorus, and then compared the two while using the TAL Chorus for the X4, and the chorus on the Juno. They're pretty close.

The problem lies in comparing the high end, I noticed the Roland has a softer edge to it. The MOTU interface I have sports a useful oscilloscope and FFT analysis, so I stripped the two tones down to the most basic saw, played a middle C, and here are the results (attached photos).

The Mopho has a more abrupt wave, more angular, with a slightly concave down shape and a small notch at the top. The Juno has a very noticable concave up shape in the bottom part of the saw. These are matched levels, with no filters applied.

Does this mean anything? Probably not, but I thought I'd contribute this to the conversation (even though you all are talking about higher end synths than these two).

[guyaguy]

It means that if someone is comparing the Rev2 to a Juno they’re comparing something close to a genuine saw to a weird saw/square hybrid.

[Razmo]

The two waveforms are clearly different, but what is more revealing would be if we saw these in a spectrum instead because the Juno waveform would probably have a different frequency spectrum than the DSI one... and with that curved bottom on the Juno I bet that there would be less higher harmonics, and maybe even more pronounced lows... this is probably also why the DSI waveform sounds brighter when the filter is completely open... to round off those abrupt edges, you would need to lower the cutoff a bit... which again returns us to the fact, that the "harshness" is simply just how the waveform sound unfiltered... just lower the cutoff a bit, and you would be good ... problem is, that if you do NOT know about this fact, you would like with all other synths assume, that the envelope amount should be set at max to get the whole spectrum included... you will just get the overly bright nature of it included as well.

[eXode]

I too think that my DSI is a little crisper than other synths; so I did a little test. Backstory: I've been trying to model some Juno 106 patches (borrowing from a friend) on my Mopho X4, and for the most part can get them dialed in. I began with debatably the easiest tone, Patch A11, Brass. A simple saw wave with some chorus. I replicated the patch sans chorus, and then compared the two while using the TAL Chorus for the X4, and the chorus on the Juno. They're pretty close.

The problem lies in comparing the high end, I noticed the Roland has a softer edge to it. The MOTU interface I have sports a useful oscilloscope and FFT analysis, so I stripped the two tones down to the most basic saw, played a middle C, and here are the results (attached photos).

The Mopho has a more abrupt wave, more angular, with a slightly concave down shape and a small notch at the top. The Juno has a very noticable concave up shape in the bottom part of the saw. These are matched levels, with no filters applied.

Does this mean anything? Probably not, but I thought I'd contribute this to the conversation (even though you all are talking about higher end synths than these two).

The two waveforms are clearly different, but what is more revealing would be if we saw these in a spectrum instead because the Juno waveform would probably have a different frequency spectrum than the DSI one... and with that curved bottom on the Juno I bet that there would be less higher harmonics, and maybe even more pronounced lows... this is probably also why the DSI waveform sounds brighter when the filter is completely open... to round off those abrupt edges, you would need to lower the cutoff a bit... which again returns us to the fact, that the "harshness" is simply just how the waveform sound unfiltered... just lower the cutoff a bit, and you would be good ... problem is, that if you do NOT know about this fact, you would like with all other synths assume, that the envelope amount should be set at max to get the whole spectrum included... you will just get the overly bright nature of it included as well.

Unless those measurements were taken directly from the waveform output those "differences" don't mean much, as there's possible coloration/distortion, etc on it's way to the output, even with the filter completely open.

The DSI products seem to have a very high frequency range for their filter, going high above human hearing, some filters like i.e. the modern Moog filters seem to cap the knob at about 16 KHz for instance (though the modulation allows it to go beyond that). On top of that the DSI products also have a very deep modulation range, again it seems to allow for deeper/wider modulation than i.e. the modern Moogs (mentioning the modern Moogs because I know them).

I believe the perceived brassiness on DSI products is because of it's wide filter range, combined with modulation depth range, making it easy to go bright compared to some other synths. The key is really in dialing back the filter frequency, as well as the mod depth. Start from very little and work your way up, carefully.

[psionic11]

Shay, don't forget that you can wave shape the Rev2 saws, and the colors available are slightly warmer or hollow.

In trying to emulate the basic osc tones this is important.  The Moog One saw shaper rounds off the highs.

Also, the Rev2 tri/saw shapes are less harsh.

As for the Curtis filter, the 2 pole responds differently and can be less harsh as well.

The filter key track acts as a secondary cutoff knob, and dialing it left really shaves the highs.

Finally, you can use filter FM or slight FX distortion to slightly change the Curtis 4 pole sonic profile.

[guyaguy]

Here are some spectro screenshot showing some of what's been described here. These images are an init sawtooth patch playing C1--no filtering. Only output volume was adjusted to match levels. You can see how the Rev2 has a bit more going on close to the 10K range compared to both the Matrixbrute and the Moog Subsequent 37. You don't always want that extra range where normally your cymbals or banshees live but it can be useful.

[guyaguy]

And here are shots of the Moog Subsequent 37 and Rev 2 filters set with resonance all the way up. The screenshot shows the frequency on the right where oscillation starts to happen. On the Moog it's just past 10K. On the Rev2 it's literally off the charts! That filter can squeal!

Notice, though, that the bass is still intact on the Rev2 here; in this case it's louder than the Moog's in fact. But it's probably the squeal that's going to catch your attention!

[MoogieBoogieMan]

Shay, don't forget that you can wave shape the Rev2 saws, and the colors available are slightly warmer or hollow.

In trying to emulate the basic osc tones this is important.  The Moog One saw shaper rounds off the highs.

Also, the Rev2 tri/saw shapes are less harsh.

As for the Curtis filter, the 2 pole responds differently and can be less harsh as well.

The filter key track acts as a secondary cutoff knob, and dialing it left really shaves the highs.

Finally, you can use filter FM or slight FX distortion to slightly change the Curtis 4 pole sonic profile.

Can you waveshape on the 08?

[Sacred Synthesis]

Can you waveshape on the 08?

No, you can't.  But there is a sawtooth-triangle combination waveform.

[MoogieBoogieMan]

Here are some spectro screenshot showing some of what's been described here. These images are an init sawtooth patch playing C1--no filtering. Only output volume was adjusted to match levels. You can see how the Rev2 has a bit more going on close to the 10K range compared to both the Matrixbrute and the Moog Subsequent 37. You don't always want that extra range where normally your cymbals or banshees live but it can be useful.

What are banshees?

[Razmo]

Here are some spectro screenshot showing some of what's been described here. These images are an init sawtooth patch playing C1--no filtering. Only output volume was adjusted to match levels. You can see how the Rev2 has a bit more going on close to the 10K range compared to both the Matrixbrute and the Moog Subsequent 37. You don't always want that extra range where normally your cymbals or banshees live but it can be useful.

What are banshees?

Undead female elves, screaming so horridly that people die if they hear them... according to fantasy role playing at least

[guyaguy]

Here are some spectro screenshot showing some of what's been described here. These images are an init sawtooth patch playing C1--no filtering. Only output volume was adjusted to match levels. You can see how the Rev2 has a bit more going on close to the 10K range compared to both the Matrixbrute and the Moog Subsequent 37. You don't always want that extra range where normally your cymbals or banshees live but it can be useful.

What are banshees?

Undead female elves, screaming so horridly that people die if they hear them... according to fantasy role playing at least

See also: Irish myths and legends 

[Jeroen van Bergen]

And here are shots of the Moog Subsequent 37 and Rev 2 filters set with resonance all the way up. The screenshot shows the frequency on the right where oscillation starts to happen. On the Moog it's just past 10K. On the Rev2 it's literally off the charts! That filter can squeal!

Notice, though, that the bass is still intact on the Rev2 here; in this case it's louder than the Moog's in fact. But it's probably the squeal that's going to catch your attention!

Thanks for sharing this! This sound character was a reason for me to get the Rev2. It can be really brutal and overwhelming.

[maxter]

I'm curious about what the cutoff is set to on an Init patch on the Prophet '08. Been quite some time since I sold mine, but if memory serves me it was 128? Or is it set to 163, like the Rev2?