0

DAC+ DSP -> DIY AVR

Fabian

August 11, 2018 07:47

None

Follow

Hello,

with the new DAC+DSP there should be SPDIF IN which should allow you to integrate your TV in your Pi music system with the benefit of DSP.

But at this point, I am not sure how one would use the DAC+DSP correctly, I would like to

 

      # have a network streaming client / spotify connect on the Pi 

      # receive my 2.0 TV Audio Signal with the DAC+DSP

      # toggle between network mode and TV mode

 

For example, if I am streaming Spotify (e.g. with Volumio) and I turn on the TV, the TV should "overrule" Spotify and the TV audio should be played. If i turn off the TV, I should be able to stream again without a reboot. This is one of the many things most AVRs can do, but it would be interesting if this would be possible with the DAC+DSP.

This leads to my questions:

 

      # how does the DAC+DSP handle the input if it gets two signals (via Pi and   SPDIF in)?

 

     # is there any software for the pi which allready allows for something like that?

 

Any comments and Inptus are appreciated! I thought it is best to ask before ordering and to use the time till the DAC+DSP is released to read up about all the details. DSP (e.g. miniDSP) is no stranger to me, but the pi world is fairly new to me - so please be gentle ;)

Thanks and regards,

Fabian

 

Edit: format is not working?

• Facebook
• Twitter
• LinkedIn

22 comments

• 1
  
  HiFiBerry team August 11, 2018 08:18

  In general the DAC+ DSP is the perfect board for your application. 

  In contrast to other DSP manufacturers, we do not limit the programs you can run on the DSP. That means you can create your own DSP program and use it with the DAC+ DSP. 
  This also means there is no "way the DSP handles signals". It simply depends on the DSP program. The DAC+ DSP comes with a program pre-installed that simply mixes SPDIF input and Raspberry Pi audio. However, it's possible to add switches for the inputs to the profile or even add an automatic switchover if a signal is detected on the SPDIF input.

  You might have a look at SigmaStudio and open one of our sample projects:
  https://github.com/hifiberry/hifiberry-dsp/tree/master/sample_files/dspproj

  This gives you a rough idea how to process data with the DSP. You can start with this and add more functionality to the project.
  With the "dsptoolkit" you can directly control a DSP program from the Raspberry Pi, e.g. changing volume control, switches or uploading filters.

  While this has a learning curve, it allows you to implement features that other DSP products can't offer.

  Best regards,
  Daniel

• 0
  
  Andreas September 15, 2018 09:10

  I just found this thread because I am looking for something similar. In my case I do not want to to have TV as input signal but a CD-Player. 

  So, it would be Raspberry Pi Audio (Squeezelite) and CD-Player and then added to a AVR and passive loudspeakers. Once there is a signal coming from the CD-Player, the input should be switched to that. Once the signal from CD-Player is off, Squeezelite should be dominant again. 

   

  So, if understood correctly, the "only" thing I need to do is create a file / map / project with this SigmaStudio software and write it onto the DSP. Is that correct?

• 0
  
  HiFiBerry team September 15, 2018 09:13

  Yes, in general that "the only thing" that you need to do. Level detectors in SigmaStudio can be a bit tricky. I recommend to have a look at our example project. The mute control in this project shows how to detect an input signal and how to control a GPIO based on this. For your project you would not switch a GPIO, but a switch in the DSP which is done in a similar way.

  Best regards,
  Daniel

• 0
  
  Andreas September 15, 2018 16:11

  Okay, will try to get familiar with SigmaStudio before I order that.

   

  One question though.... What about the Digi+ with Input and Output sockets. Would this also work with my or the thread starter´s use case ?

• 0
  
  HiFiBerry team September 15, 2018 16:14

  With the Digi+ I/O this wouldn't work as there is no simple way to detect an incoming signal.

  Best regards,
  Daniel

• 0
  
  Andreas September 15, 2018 17:13

  Hmm, just wondering as Google could not help me. SigmaStudio is currently only available for Windows, isn't it? I could not find it for MacOs.

• 0
  
  Andreas September 15, 2018 17:16

  Okay, found confirmation. Windows only. Then I have to first configure the VM. 

• 0
  
  Andreas September 15, 2018 18:07

  Good that I still had the VM with Windows 7 running. I installed it loaded one of the example projects. I looked into the LevelDetectors, Mute and I/O section. 

  
  I have to admit that I do not understand much about loudspeakers and all these specifications. And for DSPs and all the the things in that software it is even worse. So seeing all this is comparable to look at something written in Klingon (or any other alien language). 

  @Daniel 

  My and Fabian's case are nearly identical. This might be the first possibility on consumer base to combine a squeezelite (or Volumio or other linux based software) streamer with an other source (e.g. TV or CD player). Any chance that HiFiBerry could offer such a project file for all users as download?

  @Fabian

  How did you solve this? How does it turn out for you?

• 0
  
  Andreas September 19, 2018 07:00

  @HifiBerry Team,

  can you please come back with any  feedback, guidance or whatsoever?

   

   

• 0
  
  HiFiBerry team September 19, 2018 08:03

  @Andreas: Our default profile mixes SPDIF and Raspberry Pi input. We think this is a good solution for most use cases. We have no short term plans for automated switching between SPDIF and Raspberry Pi audio as different users might have different ideas what should have priority. 
  With our standard profile, you can just stop the playback on the Raspberry Pi if you want to listen to SPDIF.

  Best regards,
  Daniel

• 0
  
  Wuschel December 26, 2018 10:02

  Sorry to dig out and hijack this thread... I'm interested in using the DAC+ DSP board as well to be able to connect my Pi to my old Yamaha amp as well my CD player. The CD player should be connected to the TOSLINK-in port. Three questions regarding this:

  - How is audio from the TOSLINK input handled when it comes to bitdepth and samplerate? My CD player delivers 16bit/44.1kHz obviously. Is that passed on to the DAC as well as to the TOSLINK-out without conversion?

  - Will the audio from the Pi also be handled according to the input bitdepth and samplerate in this setup without conversion? Any will it also be output by the TOSLINK-out without conversion (if the input is 24bit/96kHz will the output be the same, switching to something different, if the source audio is different as well)?

  - Is it somehow possible to prevent the mixing of Pi audio and the TOSLINK-in port? I'd love to set it up so that TOSLINK is only used when Pi doesn't play any audio - a one-way switch, prioritizing Pi over TOSLINK-in.

• 0
  
  HiFiBerry team December 26, 2018 16:40

  Hi,

  all audio is always re-sampled to the internal sample rate of the DSP (which is 48kHz in our default profiles). You can create your own DSP profiles using SigmaStudio, if you want to use other sample rates.

  In general all kinds of features are possible, but you have to create you own DSP profiles if you want to implement a specific switching function. You might have a look at the default profiles that we provide. These are a good start for your own developments.

  https://github.com/hifiberry/hifiberry-dsp/tree/master/sample_files/dspproj

  Best regards,
  Daniel

• 0
  
  Wuschel December 27, 2018 08:45

  That's a bummer... No passthrough mode is possible to configure (with untouched audio streams from Pi and TOSLINK-in with automatic switching)? This would make the HAT perfect for a media player with an additional passthrough digital-in (and one of a kind!).

• 0
  
  HiFiBerry team December 27, 2018 09:04

  Why do you need a passthrough mode for the Raspberry and the CD player? Especially with a CD player this isn’t needed at all. The performance of the integrated sample rate converters is way better than the performance of your old Yamaha amplifier. Therefore, you won’t hear a difference.

  If you don’t want to process the audio, there is no need for a DSP. 

  Best regards,

  Daniel

• 0
  
  Wuschel December 27, 2018 10:33

  My old Yamaha amp (it's a vintage amp) only has one AUX input (the others are for record players) and it's fully analogue. Therefore the DAC+ DSP would've been the perfect solution to connect both a Pi and a CD player to the AUX-in via the DAC included in the HAT. But I'd love it to respect the incoming samplerates and bitdepths until the signal is converted by the DAC - like its done with the DAC+ Pro and Pi audio signals.

• 0
  
  HiFiBerry team December 27, 2018 11:11

  What do you expect "respecting" sample rates and bit depths will do? Basically all modern DACs use their own internal bit depth (which is 32bit for the DACs we're using). Therefore even if you just feed 16bit to it, the DAC will just add 16 zero bits to it. 
  Also using the sample rate of the source material doesn't mean the DAC performs optimal. 

  At the end, there is no such thing as "bit perfect" as all digital signals will finally be converted into the analogue domain and there will never be a DAC that can do this conversion without losses (if this makes an audible difference is a completely different question, I don't wan't to discuss here).

  Therefore re-sampling the signal to the DSP's internal sample rate is the best solution in our opinion.

  Best regards,
  Daniel

• 0
  
  Wuschel December 27, 2018 16:11

  Ok, I understand. I'm no expert in this field, but there's obviously a lot of talk out there about how optimizing audio processing is done the best way. So, I'll think and consider what the best setup for my use case is.

  The question remains whether prioritizing Pi audio over TOSLINK-in is possible to achieve and if yes, how can this be done?

  Another question: There are nice metal cases for a Pi in combination with your DAC and Digi HAT which can be bought in your shop. Is there a case for the DAC+ DSP?

• 0
  
  HiFiBerry team December 27, 2018 20:22

  Hi,

  we wouldn't recommend to use re-sampling if it would reduce audible sound quality. Modern re-samplers are working very well, especially the one in the DSP we're using here.

  In general priorizing Raspberry Pi audio over TOSLink is possible by creating a specific DSP program for this. We use level detectors for mute-control. These could also be used for this. 
  You could modify the 4way-iir-delay-inputswitcher profile for this:

  https://github.com/hifiberry/hifiberry-dsp/blob/master/sample_files/dspproj/4way-iir-delay-inputswitcher.dspproj 

  This will require using SigmaStudio to edit this and download it to the DSP.

  Best regards,
  Daniel

• 0
  
  Oscar Dubois May 26, 2020 18:10

  Dear HifiBerry support, I am very new to this...
  I am looking forward to using this DAC+ DSP for resampling my TV spdif into my big DAC I'm already happy with but my sennheisers sound a bit too neutral for my liking hence my interest in this product and no eq options on my main DAC.

  However, my question is to do with audio offset delays (due to my TV having a lip sync issue). My TV already is configured to delay audio but it maxed out at 60ms.... And I need at least 100ms or I will cringe every time I watch close-ups of people's faces and the sound is still a 20th of a second off!!! Arghhhh!!! I have done a bit of reading and saw that a DSP like for example an old Texas TPA5050 can buffer one channel by up to 170ms, as well as a certain product from miniDSP. Apologies if it's clearly in your provided HifiBerry OS software - I haven't bought anything yet - I would like to know if your HifiBerry DAC+ DSP is capable of creating an audio offset on both channels together - up to 150ms at 48kHz from the TV toslink connection?

  Thanks,

  Oscar

• 0
  
  HiFiBerry team May 26, 2020 18:33

  The DSP comes with 16kWords RAM which can held up to 1/8s (125ms) stereo at 48kHz. However, that's something you would need to create your own DSP program as the profile provided by us doesn't support long delays like this.

• 0
  
  Oscar Dubois May 29, 2020 18:49

  Thank you! I have bought everything and am successfully using Sigma Studio. I managed creating 80ms delay by inserting a couple delay blocks just before "SpdifOut1" and "SpdifOut2", modifying the program from your example "4way-iir-delay-inputswitcher" you explained here:

  https://www.hifiberry.com/docs/software/dsp-profile-example/

  It turns out that all I needed was around 75ms but that's a good thing because the software only allowed me to input up to 80ms (with the 4-way switcher program).

  Also, I read from your guide:

  "On the DAC+ DSP, only inputs 0/1 are used as it includes only a stereo DAC. Also some signals are routed to the TOSLink output."

  I am not sure if I have foolishly missed out on any filtering/equaliser processing because I reduced the values in the "4-way analogue out section" delay blocks to zero in order to create more space in the RAM...

  Secondly, I really enjoyed using the Basic Sounds modes such as "Energetic" / "Warm" when I was using the HifiBerryOS. Does the HifiBerryOS have any positive influence after I have used SigmaStudio? My concern is that you should only use the HifiberryOS if you haven't used SigmaStudio, or can they work together where I can still use the "Warm" preset? I am willing to learn how to create a "warm" EQ manually... but I really like the way you can toggle between the Basic Sounds presets so elegantly :)

  Lastly, would you recommend to use the other profiles (listed here - https://github.com/hifiberry/hifiberry-dsp/tree/master/sample_files/dspproj), such as "4way-iir-delay-mixer-asrccontrol"? Some of them seem to be updated more recently and I was wondering if I would be missing out on anything cool.

  Thanks again!

• 0
  
  HiFiBerry team May 30, 2020 05:26

  In general, HiFiBerryOS doesn't require a specific DSP profile, but can work with your own. However, the required functionalities and metadata must be present in your profile. E.g. if nor IIR filter blocks exists, no crossovers are possible. It's probably best to derive your profile from the Beocreate profile that's used in HiFiBerryOS:

  https://github.com/bang-olufsen/create/blob/master/BeocreateUniversal-v10.dspproj
  
  The important step are the metadata:. The metadata block shows HiFiBerryOS what functionalities are available and how to access these. You should also check the metadata in the universal profile to see what's expected.
  
  We don't do any work on the older profiles anymore as this is just too much work to integrate all new stuff into multiple profiles. Therefore, the profile that comes with HiFiBerryOS is our "official" one today.