Synth's Frequency Response

What determines the frequency response of a digital FM synth?

 

Well, this applies to all digital synths not just FM.
The sampling rate. It's the rate at which the audio is output. Nyquist
frequency, which is the highest frequency that can be represented in a given
sampling rate is half of the sampling rate. That sets the upper limit for
the frequency response basically. Of course the quality of the digital to
analog converters may affect the frequency response as well. Finally, all
the other elements, your sound card, speakers and mixer contrinbute to the
overall frequency response.

If you are looking for FM synthesis theory, go to:
http://www.soundonsound.com/search/

And read the first articles that come up with these searches:
1) "synth secrets part 12"
2) "synth secrets part 13"

Hope this helps.

 

I thought sampling rate only applied to digital audio (PCM, MPEG, WMA,
etc.) and not to FM synths.

Even after converting to analog, the FM signal would have to be
demodulated so that the speakers can play them. How is this done? It
is the same way an FM radio tuner eliminates the carrier wave so the
speakers can play the audio signal?

Would the "sampling rate" of the FM driver determine the frequency
response of the chip-based FM synth?

 

Thanks for the webpage.

 

FM synths are digital. There's no theoretical reason why they could not be
analog, though, other than that it would not be practical. After all, you
can simply use the output of one VCO to modulate the pitch of another and
that's called FM in the analog world (sometimes also called cross modulation
for some reason).

FM is a lot more practical digitally because it is easy to generate a number
of sine waves and implement things like different operator configurations,
patch memory etc...

Ummm, I don't get that. Aren't you talking something about FM radio
technology here? In FM synthesis, though I'm no guru, it's about sine waves
modulating each others frequencies so that extra harmonics are generated.
The result coming out is "ordinary" audio just as sampled sounds are.

You are right, the digital representation is lowpass filtered in the DAC.
Basically they remove all the components above the Nyquist that would cause
aliasing in a given sampling rate (anti-alias filter.

Aliasing is when you try to say record components that are higher than the
Nyquist. In that case, in stead of getting the frequency n Hz above the
Nyquist, you get it's alias n Hz below.

I don't know anything about the sampling rates of FM chips. I guess the rate
at which data is actually processed in the hardware determines the frequency
response (the highest frequency you can represent in the synthesis phase).
Even if some software drivers did resampling adfter the synthesis in
hardware, it would not generate extra precision, frequencies or quality to
the sound.

It's the same with digital images. If you have a 320x200 bitmap, scaling
that to 640x480 does not bring in any new detail.

 

For digital audio, the bit-resolution of the encoding determines its
dynamic range. A resolution of 1-bit gives a dynamic range of 6 dB.
This means a 16-bit wave file can accept and output loudness levels
with a difference of 96 dB. Does the same hold true for FM synth
audio?


While analog audio tends to have better bandwidth, digital audio is
less "noisy".

 

Yes, but note that the older chip-based FM synths like
OPL2 and OPL3, only have 8-bit audio even if the chip
is part of card with 16-bit wave support. I don't recall the
sample rate of the OPL chips at the moment, but it was
something like 46 kHz.


Bob Masta
dqatechATdaqartaDOTcom

D A Q A R T A
Data AcQuisition And Real-Time Analysis
www.daqarta.com

 

The FM synth signals were 8-bit?


I do believe the oldest SB cards supported 8-bit PCM at max. Never
knew about the FM synth.

Sampling rate of the wave support or FM signals?

 

FM signals. The OPL2 and OPL3 chips did _only_
FM synthesis and were completely independent
from the wave support. Their sample rate was
fixed, whereas wave sample rate was under
software control. The OPL-type operation was
continued on Creative's SB16 family, but I think
on the later ones its functionality was included
in a big Creative chip... no separate OPL chip.
But as far as I know, it still operated pretty much
the same way. The only OPL behavior that I
couldn't duplicate with the Creative chipset was
"trick" stuff like random noise generation that
was undocumented on the OPL chips.


Bob Masta
dqatechATdaqartaDOTcom

D A Q A R T A
Data AcQuisition And Real-Time Analysis
www.daqarta.com