Introducing DSPi | A powerful, user friendly and open source DSP for less than a cup of coffee

[john61ct]

Can it use interfaces' Word click as master?

Mine use the old BNC connectors to daisy chain

 

[Weeb Labs]

Can it use interfaces' Word click as master?

Mine use the old BNC connectors to daisy chain

The current I2S output implementation is a master, with slave mode planned.

The output assignment UI is progressing nicely.

 

[john61ct]

Can someone please link to the optimal case for Pico 2 W RP2350 and breakout board, want to avoid soldering

plus whatever I need to get SPDIF out

...
and for 9x analog out , ideally not RCA but as 1/4" combo female jack

allowing for TRS to output balanced, or a phono TS cable for unbalanced

I assume for consumer level amp inputs vs (+4?) dbu "pro amp" I'll need external active gain boxen?

Your link for the Pico 2 W is actually for the case. I bought this Pico 2 W

...

My case is 7x4x3 but the one you picked will likely accommodate the above.

 

[john61ct]

Different, bigger picture questions, sorry in advance for the scenario complexity, it's not just a hypothetical thought experiment. No AVP on the scene BTW.

Assume analog stereo sources only, going through a switcher into my Schiit SYN preamp,

which is in turn outputting Front LCR + Surround L/R.

However, SYN only has a mono ".1" LFE at fixed LPF, so we will let the DSPi do all the bass management.

Surround L/R pair is plain 3-ways, internal xover.

Main L/R front is "assembled active 4-way":

LS50s for above mid-bass only

+ colocated subs (not too deep, maybe stereo)

+ MBM couplers bandpassed in between, def stereo HPF ~80Hz

plus for centre, maybe FR + big/deep trueSub

plus one or more mono big/deep trueSub able to place freely for room modes.

Whew!

Say 9 channels is not enough, can I coordinate multiple rPi DSPi units as one system?

What's the best way to get the five SYN analog RCA output channels into the rPi's inputs?

If possible please link to USA sources?

Does this mean my DSPi case needs to gets bigger? or do I split to multiple boxen?

 

[Rocky Maine]

Can someone please link to the optimal case for Pico 2 W RP2350 and breakout board, want to avoid soldering

plus whatever I need to get SPDIF out

...
and for 9x analog out , ideally not RCA but as 1/4" combo female jack

allowing for TRS to output balanced, or a phono TS cable for unbalanced

I assume for consumer level amp inputs vs (+4?) dbu "pro amp" I'll need external active gain boxen?

Breakout board SPDIF board

My case is 7x4x3 but one could use something smaller

Can't help with the other requirements you state

My use case is quite simple, mainly room eq at low levels, aka "loudness"

 

[Wesseling Audio Lab]

I now have USB selective suspend disabled. No change. Dropout twice in first 30 minutes today.

Streaming from ipad now to see what happens

 

[Weeb Labs]

I now have USB selective suspend disabled. No change. Dropout twice in first 30 minutes today.

Streaming from ipad now to see what happens

Could you please describe the sound of your dropouts? Are they clicks or simply a very short gap in output?

 

[Wesseling Audio Lab]

Could you please describe the sound of your dropouts? Are they clicks or simply a very short gap in output?

They're very short gaps in output! A few hours with ipad now and not a hint of dropout (with the same usb cable to DSPi and certified charge adapter)

 

[Weeb Labs]

They're very short gaps in output! A few hours with ipad now and not a hint of dropout (with the same usb cable and certified charge adapter)

I've implemented some additional statistics while I work on the logging system. This measures the time since the last dropout event and time between the last two.

This is under macOS and as you can see, it ran for almost six hours (3 before first event) without a dropout. Interestingly, the DMA starvation counter seems to reset during each event, which may offer a clue. As part of the logging, I will be adding an uptime counter.

The host appears to be triggering stream restarts, which would explain why the DMA starvation increments but not the underrun counter. The latter is disabled during stream mode changes.

I believe the problem may be our feedback servo. It was never revised after the new buffer architecture, which means it is now several times stronger than it should be. This is the first issue that needs to be addressed.

 

[Red_Red]

Different, bigger picture questions, sorry in advance for the scenario complexity, it's not just a hypothetical thought experiment. No AVP on the scene BTW.

Assume analog stereo sources only, going through a switcher into my Schiit SYN preamp,

which is in turn outputting Front LCR + Surround L/R.

However, SYN only has a mono ".1" LFE at fixed LPF, so we will let the DSPi do all the bass management.

Surround L/R pair is plain 3-ways, internal xover.

Main L/R front is "assembled active 4-way":

LS50s for above mid-bass only

+ colocated subs (not too deep, maybe stereo)

+ MBM couplers bandpassed in between, def stereo HPF ~80Hz

plus for centre, maybe FR + big/deep trueSub

plus one or more mono big/deep trueSub able to place freely for room modes.

Whew!

Say 9 channels is not enough, can I coordinate multiple rPi DSPi units as one system?

What's the best way to get the five SYN analog RCA output channels into the rPi's inputs?

If possible please link to USA sources?

Does this mean my DSPi case needs to gets bigger? or do I split to multiple boxen?

I'd like to try and answer your questions, but I don't really see the point of DSPi in your setup... How do you think DSPi could be useful for you?

 

[john61ct]

Applying DSP filters (the convolving part) without having to dedicate an expensive power thirsty PC to staying on all the time.

Getting from analog input (especially balanced) to USB input to feed the rPi seems hard, or at least very costly - I guess each 2-ch input pair needing a separate ADC unit, so I'd need three.

Having three separate DSPi units seems hard to coordinate.

I'm finding apparently crappy devices designed to record cassette / VHS audio, not much cheaper.

Is there really no other SQ path to get channels into DSPi?

If only the PocketADC had USB out, at $55/unit would be a bargain.

If DSPi accepted S/PDIF input looks like that would be the way to go.

Is it the design intention to feed signals to the DSPi only from another computer? If so I don't see the point, just have the player PC handle the filtering!

I hope I've just missed something...

 

[Red_Red]

Applying DSP filters (the convolving part) without having to dedicate an expensive power thirsty PC to staying on all the time.

Getting from analog input (especially balanced) to USB input to feed the rPi seems hard, or at least very costly - I guess each 2-ch input pair needing a separate ADC unit, so I'd need three.

Having three separate DSPi units seems hard to coordinate.

I'm finding apparently crappy devices designed to record cassette / VHS audio, not much cheaper.

Is there really no other SQ path to get channels into DSPi?

If only the PocketADC had USB out, at $55/unit would be a bargain.

If DSPi accepted S/PDIF input looks like that would be the way to go.

Is it the design intention to feed signals to the DSPi only from another computer? If so I don't see the point, just have the player PC handle the filtering!

I hope I've just missed something...

No, you're not mistaken, but it will indeed be expensive (ADC + DACS)... However, it does plan to have at least two SPDIF or I2S inputs for the DSPi (in addition to USB). But not simultaneously... well, yes and no. It should be possible to mix two inputs together... but not to route all inputs to the outputs simultaneously (in any case, this scenario hasn't been documented by its author).
But otherwise, upstream of your SYN you can always use the DSPi to do quite a lot of things...

 

[john61ct]

My primary need and certainly the first to tackle, is per speaker compensation and crossovers, especially time / phase issues. I may end up not using DSP at all if it is not necessary for that.

 

[capslock]

I've implemented some additional statistics while I work on the logging system. This measures the time since the last dropout event and time between the last two.

View attachment 520477

This is under macOS and as you can see, it ran for almost six hours (3 before first event) without a dropout. Interestingly, the DMA starvation counter seems to reset during each event, which may offer a clue. As part of the logging, I will be adding an uptime counter.

The host appears to be triggering stream restarts, which would explain why the DMA starvation increments but not the underrun counter. The latter is disabled during stream mode changes.

I believe the problem may be our feedback servo. It was never revised after the new buffer architecture, which means it is now several times stronger than it should be. This is the first issue that needs to be addressed.

I haven't gotten around to setting up an DSPi, but let me share my experience with Chromecast streaming in my Fritz MESH WLAN network (and with MESH disabled, I get similar results):
- original Chromecast Audio has occasional drop-outs or stutter when streaming CD-quality (16/44), worse for 96 kHz, no matter whether I use Tidal, Qobuz, Deezer or Amazon music
- Google Nest Audio is fine with 16/44 though I haven't gotten a stereo pair to work, so this is mono only
- Wiim Amp is fine both through Chromecast protocoll or native streaming of Tidal or Amazon Music up to 24/192, but it takes noticably longer to buffer (i.e. changing a track takes a few seconds)

So buffer size may matter, too.

 

[john61ct]

I wonder if this ADC gizmo will work to deliver stereo line in to DSPi ?

Cubilux HLMS-C5 $26
USB Type C interface, bus powered
6.35mm 1/4" Stereo Line-In
ADC 48KHz, 16/24-bit

driverless so UAC?, but I see some CYA about "PC or laptop not phones or tablets" so likely not UAC2

 

[john61ct]

Otherwise Behringer U-Phoria UMCx0xHD at double the price per channel

Focusrite Scarlett 2i2 is double that again

 

[john61ct]

it does plan to have at least two SPDIF or I2S inputs for the DSPi

2x S/PDIF would be fine, do not need concurrent with other paths.

Then a miniDSP PocketADC would give known-good ADC service, compact form factor.

Question remains, with 3x DSPi to handle the 6x channels, is there going to be any way to coordinate timing? or say crossover FR settings?

Otherwise, I guess using REW to measure gives the bird's eye overview for feedback...

 

[Wesseling Audio Lab]

I have considered that dropout could be from streaming service running wireless WLAN, although I experience dropout with windows and PC is connected via ethernet connection.
No dropout with IPAD streaming with WLAN

 

[somebodyelse]

I wonder if this ADC gizmo will work to deliver stereo line in to DSPi ?

Cubilux HLMS-C5 $26
USB Type C interface, bus powered
6.35mm 1/4" Stereo Line-In
ADC 48KHz, 16/24-bit

driverless so UAC?, but I see some CYA about "PC or laptop not phones or tablets" so likely not UAC2

Not directly. It needs a USB host to work, but the DSPi is not a USB host. You would have to connect both of them to something like a PC, Pi or other computer and use that to route the audio from one to the other.

The MiniDSP PocketADC should work some time in the future when the planned spdif input is implemented.

 

[somebodyelse]

Otherwise Behringer U-Phoria UMCx0xHD at double the price per channel

Focusrite Scarlett 2i2 is double that again

Same problem as the Cubilux - you need a USB host.