Subwoofer LEDS – RGB, brightness controlled by amp

A MOSFET will be a good choice.
The following gives a guide to directions - ask more questions.

Load max sounds like (pun notice retrospectively) LED strips at 12v, 10A
MOSFET needs to switch this load so 20V+ rating desired, usefully more than 10A - say 20A+.

Now the fun begins.
If the FET is hard on then dissipation is low. With and Rdson (= on resistance) of say 10 milliOhm the power loss at 10A due to channel resistance = I^2R = 10^2 x 0.01 = 1 Watt.
If you PWM switch the FET you add switching losses but losses are still "moderate"
If you linearly switch the FET so it acts like a resistor dissipation is much higher.
A 12V LED strip will decrease to close to 0% current by 8V (or higher) probably, so if full load is 10A you will probably be down to 5A at 10V (so driver loss = (12-10) * 5 = 10 Watts, and down to say 1A at 8V (loss = (12-8)*1 = 4 Watts.
So you can probably expect losses in linear mode to be under say 10W but allowing for 20W may be safe. Rdson becomes non-critical when used this way as resistive on losses swamp Rdson.

So, we have a MOSFET rated at ~~~= SOME OF 20V, >10A and 20W actual dissipation and Rdson = 10 milliOhm.
I say "some of" as you don't need Rdson low in linear mode but you don't need as hogh dissipation in PWM mode.
For 1 off amateur use specify leaded TO220 package for easier use and ease of heatsinking.

Searching DIGIKEY with
MOSFET 30V 20A TO220
then sorting by ascending price and looking for lowest $ in stock 1 quantity parts gives as below. I didn't specify Rdson but could have.

Infineon PSMN022-30PL - datasheet here in stock at 73 cents US in 1's. Pricing
30V, 30A, TO220, Rdson 34 milliOhm max at 5V gate drive, 41 W max dissipation,

See fig 1 for max dissipation based on heatsink temperature.
Thermal resistance is 3.6 C/W junction to case so you'd want to keep dissipation under 20W and ideally under 10W - SO probably OK.

So - how do you drive this.

• You MAY be able to drive it straight from the amp with a potentiometer to adjust level. MOSFET in linear mode, probably.

• Feed amp output to a comparator to make rail-rail signal and drive FET. Lower dissipation. Don't exceed max gate drive. May not look like what you want to see. May. MOSFET in PWM mode - low losses.

• Use a linear amplifier to adjust level up or down as required. Can add simple filtering to adjust hang time, attack time, frequency response, more ... . These are all "easy enough" but lets see if this is what you want. MOSFET in linear mode.

... 30PL - slightly better Rdson, slightly more $

Wow!
Infineon IPP065N03LGIN - pricing and datasheet
30V, 50A, 6.5 milliOhm Rdson, 2.7 C/W, TO220 10 milliOhm Rdson worst case at gate drive of 4.5V. $US1.05/1 in stock.
Looks very good for this task.

Costs of COB, RGB and phosphor

Those particular LEDs are individually addressible, and that brings a host of efficiency problems along.

• The on-LED controller needs its own power.
• Supply power is not a voltage that is ideal for LEDs - the LEDs want ~3V but the supply is ~5V. That's most likely a linear driver (I really doubt that chip includes the chokes for 3 buck converters), you are chopping 5V down to 3V linearly, discarding 40% of the energy before it even gets to the emitter proper.
• Power distribution is at only 5V, which is extremely vulnerable to transmission losses. I've seen designs that called for 350kcmil + and - buses. You gotta be kidding.

The upshot is that individually addressable LEDs are optimized for building billboards and stadium TVs, not illumination.

Now if you want to go for the non-addressible RGB LED strips, they mostly cure these problems. No controller to waste power. Distribution is at 12V (somewhat better) or 24V (much better). 12V is better for powering a stack of 3 LEDs, you maybe lose 20% to the limiting resistor.

That said, the notchy R/G/B light is not going to look quite right to everyone, and won't have an excellent CRI. It's not white light, it's just designed to trick the eye into looking white. I would expect it to give the same kind of eye fatigue you get from looking at a screen for a long time.

On the upside, since the LED is internally doped to make that specific color, it's not using a phosphor to make the color, so it's not paying phosphor losses.

Now if you go with single-color white LEDs, those actually make blue internally and use phosphors to convert the blue into a full spectrum. Phosphors have an efficiency loss (and they cost money). That's why bluer light LEDs (6000k) are cheaper and more efficient than less blue LEDs (2700k).

Don't forget aiming

There's one more angle (heh) to efficiency. An old bulb emits spherical light. It even illuminates the bottom of its own socket. But nobody wants to light a sphere. They typically want a cone or wedge ranging from 8-120 degrees. Historically you solved that with inefficient reflectors.

LEDs are the first to give a cone of light, of roughly 140-160 degrees. And that lends itself to optical (lens) aiming, which is near 100% efficient. Now you don't have to light up the wall, the sky, the neighbor's bedroom window - you can just light the subject. That is a gigantic efficiency advantage that you should make full use of.