MiniDSP Flex HTx

[masterhw]

Obviously this all has to be downmixed to at most 7.1.

And yes -- no height channels.

This is what I meant - it’s not that the sound didn’t play, the sounds begin life as a 7.1 mix. The metadata to fold it out into height is not used.

HDMI supports up to 32 LPCM channels or something, and I’ve thus far been able to find any example of Dolby content outputting to more than 8 (7.1). I think it must be a licensing thing.

 

[rirelien]

Note: I've responded in the hopes of sharing info and being corrected where I'm wrong. Take this as inquiry!

Applications like Apple TV+ and YouTube are all lossy, so that is lossy audio converted to LPCM.

In my understanding -- yes, agreed.

Plex is like I mentioned (along with Infuse), a third party app that supports actual DTS-HD/TrueHD decoding to LPCM.

Right. I do use Plex for this purpose, on ATV.

Apple Music is interesting, though. As far as we know, it is legitimately outputting 24bit 48kHz (for stereo music). Given that Apple Music is a native Apple application, that indicates that Apple doesn’t see the ATV as just a lossy platform (and why I think it is misleading to talk about it like it is). No reason to think of it as such.

Again, agreed, in principle. What the platform (hardware + software) supports is distinct from what applications on that platform choose to support.

I would say that the platform definitely supports outputting up to 7.1 LPCM (at what bit depths/sample rate max?); whether your source is lossy or lossless is an application-level concern.

The OS platform claims to support both E-AC3 and TrueHD streams with its AVPlayer framework; this in theory lets the system decide whether to pass through the stream or decode it onboard to LPCM.

Apps like Infuse probably use their own player, bypass AVPlayer, etc.

Atmos in Apple Music and the streaming apps are all lossy E-AC3. The spatial metadata isn’t discarded because ATV4K supports lossy Atmos, just not lossless Atmos. Unless you’re talking about using the ATV4K with the Flex HTx because then yes, the metadata is discarded.

My point here is that the metadata is not discarded -- it's used to render the 5.1 or 7.1 output. It's NOT passed along, because LPCM wouldn't support it anyway! To pass it along, you'd need the output to be inside a container that supports this metadata, which is passed to an external Atmos renderer (ie in your receiver/processor.) This is not my use case so I haven't spent time on this, but my understanding is that this works for lossy Atmos but no one has found a way for this to work for lossless Atmos formats from a third-party player -- not sure if that's right?

When the metadata is used to render object-based spatial audio to 5.1 or 7.1, clearly there's no way of rendering distinct height channels with that channel layout -- but you get those objects rendered with what's available. In other words, I hear height objects still.

Handy test tracks (lossy):

(Potentially offtopic, but: I have used APL Virtuoso with Apple Music on a Mac, and can see the Atmos renderer there sending the height data as distinct channels.)

 

[rzrike]

Note: I've responded in the hopes of sharing info and being corrected where I'm wrong. Take this as inquiry!


In my understanding -- yes, agreed.


Right. I do use Plex for this purpose, on ATV.


Again, agreed, in principle. What the platform (hardware + software) supports is distinct from what applications on that platform choose to support.

I would say that the platform definitely supports outputting up to 7.1 LPCM (at what bit depths/sample rate max?); whether your source is lossy or lossless is an application-level concern.

The OS platform claims to support both E-AC3 and TrueHD streams with its AVPlayer framework; this in theory lets the system decide whether to pass through the stream or decode it onboard to LPCM.

Apps like Infuse probably use their own player, bypass AVPlayer, etc.


My point here is that the metadata is not discarded -- it's used to render the 5.1 or 7.1 output. It's NOT passed along, because LPCM wouldn't support it anyway! To pass it along, you'd need the output to be inside a container that supports this metadata, which is passed to an external Atmos renderer (ie in your receiver/processor.) This is not my use case so I haven't spent time on this, but my understanding is that this works for lossy Atmos but no one has found a way for this to work for lossless Atmos formats from a third-party player -- not sure if that's right?

When the metadata is used to render object-based spatial audio to 5.1 or 7.1, clearly there's no way of rendering distinct height channels with that channel layout -- but you get those objects rendered with what's available. In other words, I hear height objects still.

Handy test tracks (lossy):

(Potentially offtopic, but: I have used APL Virtuoso with Apple Music on a Mac, and can see the Atmos renderer there sending the height data as distinct channels.)

Lossless Atmos has a fallback 7.1 TrueHD included with it, so if you discard the Atmos metadata, you still have 7.1 TrueHD with only the loss of height channels. Rip a 4K blu-ray to mkv and throw the audio in Audition, you’ll see eight channels.

When it comes to lossy Atmos, the ATV4K still doesn’t really do passthrough. It includes the Atmos metadata in a Dolby MAT container (not something I fully understand) alongside the LPCM. This currently only works with lossy Atmos.

 

[deercreekaudio]

Here is our video that overviews the HTx archetchure:

 

[andrewtungnguyen]

For anyone considering MiniDSP Flex HTx with a Raspberry Pi: once you move bass management out of the AVR and let MiniDSP handle crossover, delays, and routing, you get capabilities most AVRs simply can’t do. The main advantage is true stereo bass (L→Left sub, R→Right sub) with independent delay and PEQ alignment, instead of the usual L+R mono sub channel that all AVRs sum. This allows phase-accurate integration with the mains, adjustable down to tenths of a millisecond, and lets you choose any crossover slope and point without being confined to the simple filters in an AVR.
Using the Pi, you can automate preset switching based on AVR input or content type (ex: Music, Cinema, EDM, Infra), and automatically load BEQ profiles per movie without manual intervention. It also allows different PEQ structures per preset while maintaining level matching and delay integrity.
The result is not about louder bass — it’s about accurate integration: localization improves, mid-bass transient response tightens, and mains + subs behave like a single system instead of separate components. It functions more like a studio-style active crossover rather than conventional home-theater bass management.
Flex HTx basically provides the tools AVR bass management lacks, and the Pi provides automation to make it easy to live with.

 

[markstein]

The main advantage is true stereo bass (L→Left sub, R→Right sub) with independent delay and PEQ alignment, instead of the usual L+R mono sub channel that all AVRs sum. This allows phase-accurate integration with the mains, adjustable down to tenths of a millisecond, and lets you choose any crossover slope and point without being confined to the simple filters in an AVR.

Why would one want that?

A Linkwitz Riley x-over at 80 Hz and mono bass, using multiple subs and MSO optimization, works pretty well.

 

[andrewtungnguyen]

Why would one want that?

A Linkwitz Riley x-over at 80 Hz and mono bass, using multiple subs and MSO optimization, works pretty well.

When you move bass management out of the AVR and into the HTx, you can route L to the left sub and R to the right sub with independent delay, gain, and PEQ instead of the AVR’s mono summed sub channel. That lets you time-align each side down to tenths of a millisecond and fix side-specific room issues instead of averaging them together. It’s not a stereo bass imaging thing — it’s just tighter integration and cleaner crossover handoff with the mains.
And yeah, four subs with MSO can do great things if you have perfect placement options, but not everyone has the space or flexibility. Two subs treated like part of the speaker system can be a simpler path that still gives very accurate results.
Using the Pi also lets you automate preset switching and BEQ loading per movie or music source, so gain and delay integrity stay intact without touching menus all the time. Basically the Pi handles the busywork and the HTx handles precision.
Not trying to argue it’s better for everyone, just a different way to solve the problem depending on your room and goals.

 

[markstein]

Well, at 80 Hz and below in typical (European) listening rooms I would say there's no "left" and "right" as you are in the modal region of the room. So I personally do not care about that.

The super exact alignment you are describing might look appealing on paper, however, I doubt there is any perceivable benefit. On the contrary - a left sub / right sub config leads to worse results than, e.g. a front sub / back sub. You may want to study Todd Welti's work on multi subwoofer placement in typical small rooms.

 

[dweeeeb2]

@andrewtungnguyen how is the pi doing the switching? ie is the HTx changing presets and PEQ so Pi acting like fancy remote, or is EQ occurring in the pi?

 

[andrewtungnguyen]

Well, at 80 Hz and below in typical (European) listening rooms I would say there's no "left" and "right" as you are in the modal region of the room. So I personally do not care about that.

The super exact alignment you are describing might look appealing on paper, however, I doubt there is any perceivable benefit. On the contrary - a left sub / right sub config leads to worse results than, e.g. a front sub / back sub. You may want to study Todd Welti's work on multi subwoofer placement in typical small rooms.

Totally fair points. I’m not chasing left/right bass localization at ~80 Hz it’s mostly modal anyway. I’m just experimenting because in my setup the mains are literally stacked right on top of each sub, so they share almost the same acoustic origin. Using the HTx just lets me dial in delay/phase per side to clean up the crossover handoff at the MLP. Center/surround bass will still be mono to both subs, so it’s not pure stereo bass.

 

[andrewtungnguyen]

@andrewtungnguyen how is the pi doing the switching? ie is the HTx changing presets and PEQ so Pi acting like fancy remote, or is EQ occurring in the pi?

The Pi isn’t doing any EQ or DSP. All filters/delays/PEQ/crossover live inside my MiniDSP (I just ordered the HTx, so I already realize what it's capable of since I'm doing basically the same with just a 2x4HD it will just take more time to change code). The Pi just automates preset switching and BEQ loading by watching AVR input state and Plex webhooks. Basically a smart remote that handles the busywork.

Here’s a simplified slice of what it’s doing:
Copy code
Python
import requests
from flask import Flask, request

def load_preset(preset_number: int):
requests.post(f"{MINIDSP}/minidsp/preset/{preset_number}")

def unload_beq():
requests.post(f"{BEQ}/unload")

def apply_beq(beq_id: str):
requests.post(f"{BEQ}/apply", json={"id": beq_id})

def search_beq(title: str):
r = requests.get(f"{BEQ}/search", params={"q": title})
results = r.json().get("results", [])
return results[0]["id"] if results else None
Example logic when AVR input changes:
Copy code
Python
def on_avr_input_change(source: str):
if source == "WIIM":
unload_beq()
load_preset(0) # Music Ref
elif source == "FIRETV":
load_preset(1) # Cinema
elif source == "XBOX":
load_preset(2) # Game / EDM
And example Plex webhook handler:
Copy code
Python
app = Flask(__name__)

@app.route("/plex", methods=["POST"])
def plex_webhook():
payload = request.json
title = payload["Metadata"]["title"]
beq_id = search_beq(title)
if beq_id:
apply_beq(beq_id)
return "ok"

So yeah — Pi = automation + preset switching + BEQ.
MiniDSP / HTx = actual audio processing.
Also lets me automate different crossover architectures per content type without touching anything. The Pi also converts subwoofer output data into an envelope that drives the Govee lights as a visualizer

 

[EJ3]

Totally fair points. I’m not chasing left/right bass localization at ~80 Hz it’s mostly modal anyway. I’m just experimenting because in my setup the mains are literally stacked right on top of each sub, so they share almost the same acoustic origin. Using the HTx just lets me dial in delay/phase per side to clean up the crossover handoff at the MLP. Center/surround bass will still be mono to both subs, so it’s not pure stereo bass.

My mains are also stacked: on top of my floor firing subs. And I do not have an AVR. (I think that this device is something to look into after the first 1/4 of next year, it seems that it may cause me to get a TV again [something my wife & I have not had since 2007]) to watch Blue Rays & 4K disks on.

 

[andrewtungnguyen]

Honestly I don’t think I’m crazy. With proper time alignment and active XO I can feel bass energy move across the screen when sounds pan left to right. Not deep sub bass, but the crossover region around 60 to 100 Hz follows the soundstage. There is research showing humans can localize in that range under good conditions, and high end speakers like KEF Blade, Perlisten S7t and Kii BXT are designed around the same idea of paired full range and low frequency drivers. So I doubt I am the only one who has experienced it.

 

[EJ3]

Honestly I don’t think I’m crazy. With proper time alignment and active XO I can feel bass energy move across the screen when sounds pan left to right. Not deep sub bass, but the crossover region around 60 to 100 Hz follows the soundstage. There is research showing humans can localize in that range under good conditions, and high end speakers like KEF Blade, Perlisten S7t and Kii BXT are designed around the same idea of paired full range and low frequency drivers. So I doubt I am the only one who has experienced it.

My home made subs (a pair) raw speakers (automotive competition dual 4 OHM voice coil 12" Pioneer [re-circuited to appear as a single 4 OHM voice coil to my amps]) have a FR of 20 Hz-80 Hz. My mains have an FR of 26 Hz-20 KHz +-2 Db (without any type of EQ).
Measured:
The close-miked woofer (and port) response was also considerably flatter than we have measured from most speakers, with a very small bass-resonance peak. At the system resonance of 60 Hz, the output was only about 2 dB above its average level in the upper part of the woofer’s range, and even that minor output variation was spread over almost two octaves. When the bass curve was spliced to the room-response measurement, the resulting composite frequency response was flat within about ±2 dB from 26 to 20,000 Hz. The horizontal directivity of the tweeter was only discernible in the room measurement above 10,000 Hz.
Quasi-anechoic FFT measurements showed an overall group-delay variation of about 0.1 millisecond between 4,000 and 20,000 Hz and 0.5 MS. between 1,000 and 20.000 Hz, convincing evidence of the attention paid to the phase characteristics of the M-905.
So, with them sitting directly on top of the floor firing subs (port tuned to 29 Hz [and with over 1 KW at 4 OHMs per sub]), suspect that the MiniDSP Flex HTx would be a great improvement.

 

[andrewtungnguyen]

My home made subs (a pair) raw speakers (automotive competition dual 4 OHM voice coil 12" Pioneer [re-circuited to appear as a single 4 OHM voice coil to my amps]) have a FR of 20 Hz-80 Hz. My mains have an FR of 26 Hz-20 KHz +-2 Db (without any type of EQ).
Measured:
The close-miked woofer (and port) response was also considerably flatter than we have measured from most speakers, with a very small bass-resonance peak. At the system resonance of 60 Hz, the output was only about 2 dB above its average level in the upper part of the woofer’s range, and even that minor output variation was spread over almost two octaves. When the bass curve was spliced to the room-response measurement, the resulting composite frequency response was flat within about ±2 dB from 26 to 20,000 Hz. The horizontal directivity of the tweeter was only discernible in the room measurement above 10,000 Hz.
Quasi-anechoic FFT measurements showed an overall group-delay variation of about 0.1 millisecond between 4,000 and 20,000 Hz and 0.5 MS. between 1,000 and 20.000 Hz, convincing evidence of the attention paid to the phase characteristics of the M-905.
So, with them sitting directly on top of the floor firing subs (port tuned to 29 Hz [and with over 1 KW at 4 OHMs per sub]), suspect that the MiniDSP Flex HTx would be a great improvement.

Yeah I think it would easily elevate your clever setup ... It sounds like you already have useful measurements that you could transfer directly and run sweeps.

 

[EJ3]

Yeah I think it would easily elevate your clever setup ... It sounds like you already have useful measurements that you could transfer directly and run sweeps.

I hadn't particularly thought about it being clever, when I first did it years ago, it was due to "where to put things" & them still be useful to me. It worked well enough that, no matter where I was, I used the same setup.
The system has been with me since the 1990's & moved from James Island, SC, to Saipan, then to Guam, to Johns Island, SC and now back in the Tri County area around Charleston, SC.
These moves have hit the upgrade budget, but I think that I'll be able to pull that one up after the first 1/4 of 26.
It will be a heck of a learning curve for me, but the probable benefits seem enormous.
Thanks for the positive encouragement.

 

[andrewtungnguyen]

I hadn't particularly thought about it being clever, when I first did it years ago, it was due to "where to put things" & them still be useful to me. It worked well enough that, no matter where I was, I used the same setup.
The system has been with me since the 1990's & moved from James Island, SC, to Saipan, then to Guam, to Johns Island, SC and now back in the Tri County area around Charleston, SC.
These moves have hit the upgrade budget, but I think that I'll be able to pull that one up after the first 1/4 of 26.
It will be a heck of a learning curve for me, but the probable benefits seem enormous.
Thanks for the positive encouragement.

If you are just doing summed bass and crossover without need for more than 4 channels I think minidsp 2x4 hd or DDRC 24 would be sufficient btw . Im just getting flex htx because I ran out of headroom trying to cross my bookshelf's and two large subs

 

[EJ3]

If you are just doing summed bass and crossover without need for more than 4 channels I think minidsp 2x4 hd or DDRC 24 would be sufficient btw . Im just getting flex htx because I ran out of headroom trying to cross my bookshelf's and two large subs

The same thing that I will be trying to do. They are about 7.5 feet apart & I can move them about 2 feet each left one left & right one right (toward the side walls and about 1 & 1/2 feet out from the back wall. The ports on the main speakers are forward firing & the subs & ports are downward firing. As I run them now, each (of 2) sub amp is receiving a summed to mono signal, I want to experiment with stereo subs, placement (within what I can do). The home décor specialist (wife) says that I can get rid of some furniture, if I want to. But our place is small & if I deem it's in my way, it has to find another home. So I am hesitant to do that because it's a permanent deal. Also, she is the one that bought the furniture for that room (the living room). But bought more than we needed, due to pressure from my mother (who has a huge home).

 

[andrewtungnguyen]

The same thing that I will be trying to do. They are about 7.5 feet apart & I can move them about 2 feet each left one left & right one right (toward the side walls and about 1 & 1/2 feet out from the back wall. The ports on the main speakers are forward firing & the subs & ports are downward firing. As I run them now, each (of 2) sub amp is receiving a summed to mono signal, I want to experiment with stereo subs, placement (within what I can do). The home décor specialist (wife) says that I can get rid of some furniture, if I want to. But our place is small & if I deem it's in my way, it has to find another home. So I am hesitant to do that because it's a permanent deal. Also, she is the one that bought the furniture for that room (the living room). But bought more than we needed, due to pressure from my mother (who has a huge home).

I use to actually have four subs so I am familiar with MSO except I just manually aligned them using 2x4 HD. Eventually home decor specialist found them to be an eyesore so I had to find new homes for two subs. I was just sitting there thinking I have extra channels maybe I'll try hooking my mains to it which led me down this path. It one hundred percent worked but it was asking too much of the 2x4 HD at higher volumes. I hope Flex HTx will be the answer because it's honestly more $$$ than I paid for any of my equipment. I've always only bought used items but this one seems worth it from what I can tell. I see it as like a Trinnov mini. If Trinnov is a Porsche the HTx is a Corvette C8

 

[EJ3]

I use to actually have four subs so I am familiar with MSO except I just manually aligned them using 2x4 HD. Eventually home decor specialist found them to be an eyesore so I had to find new homes for two subs. I was just sitting there thinking I have extra channels maybe I'll try hooking my mains to it which led me down this path. It one hundred percent worked but it was asking too much of the 2x4 HD at higher volumes. I hope Flex HTx will be the answer because it's honestly more $$$ than I paid for any of my equipment. I've always only bought used items but this one seems worth it from what I can tell. I see it as like a Trinnov mini. If Trinnov is a Porsche the HTx is a Corvette C8

I am running one 95 SINAD 500+ watt a channel (@4 OHMs) amp in stereo on the mains and one of the same amps bridged mono for 1000+ watts (@ 4 OHMs) for each of my 2 subs (Amirm tested one of my 6):
Lab Input Measurements
I was surprised that the frequency response was not flat but was relieved to see later in the thread that this is due to insertion of low and high pass filters. So here is the frequency response with Lab input that doesn't have such a filter:

Response now (in green) as it should be, ruler flat to below 10 Hz, and well extending past the 40 kHz limit of this measurement.

I figured the filters may be adding some noise/distortion so re-ran the dashboard again:

Distortion doesn't change but if you look at the noise floor at 20 Hz, it is down by some 10 dB. That improves SINAD a couple of dBs, making the amplifier stand out even more!

And signal to noise ratio:

Conclusions
Nice to see innovation like this from equipment that is over 30 years old! Shame on manufacturers that produce amplifiers for much less power, more distortion and higher prices these days. No, you don't get a fancy case here and sheet metal is strictly budget category. But you are not going to sit on the amp. The guts are where it matters and NAD 2200 delivers.

NOTE: the output relay on stock 2200 gets corroded and fails over time. There are videos and DIY threads on how to upgrade the relay there to fix the problem. The unit tested here has that fix. Other than that, there are not reports of many other reliability issues even though NAD products are often said to be less reliable than other brands.

Overall, I am happy to recommend the NAD 2200. I almost gave it the highest honors but given the upgraded nature of the test unit, and the fact that used amps may have issues, I avoided that. But you could have easily pushed me to give it the golfing panther.

(EJ3 Says): The price of the MiiDSP Flex HTx is stout (especially with the added DIRAC stuff) [that DIRAC stuff may cause me to have to wait until the end of the 2nd 1/4 of 26).
But this appears to be the best way forward to better sound at the moment.