Apache Labs Community Forum

Posted: **Sun Dec 22, 2019 12:28 am**

Hi,

I know that some of you are using Digital Audio Workstations or VST Hosts for audio processing which prompted these questions:

1). I assume that you have a "dedicated" PC to run both the SDR and DAW or VST Host with plug-ins, correct?
2). If the answer to the above is in the affirmative, have you noticed audio drop-outs or other such "glitches" in your transmissions that would be traceable to using DAW or VST Host for your audio processing?
3). There are some excellent tutorials on how to build a PC that is truly dedicated to audio processing when employing a DAW or VST Host which have done away with a lot of the "bloatware" PCs come with that are deleterious on DAW and VST Host processing of audio. Depending on the extent to which one uses DAW (and especially the number and type of plug-ins) for audio processing a truly dedicated PC to audio may not be necessary, right?
4). Doesn't ANAN SDR in essence also process audio in the digital domain and you may also experience audio drop-outs, correct? The hardware in ANAN SDR with its FPGAs may make audio processing by the SDR rig inherently more efficient as well and audio drop-outs may not really be a concern at all, correct? I must admit that I don't know at all how audio is processed in ANAN and the FPGA may have absolutely nothing to do with that aspect at all.
5). Am I correct in thinking that even if one has a workstation grade machine with - say - an i9 CPU in it one might experience audio drop-outs as there are many functions such as automatic software updates and a myriad notifications that are going on in the "background" that potentially cause audio drop-outs, correct? When I say audio "drop-outs," I am mainly alluding to those that you may experience when using DAW or VST Host for audio processing.
6). Are there functions in a very hight performance PC when running just SDR that you have had to disable completely to not have any sort of audio drop-outs or other types of "glitches" when NOT using DAW or VST Host for audio processing and using what ANAN enables by itself?

It would be great to hear your thoughts and comments on the above questions.

Best,
Juha
- NI2M

Posted: **Sun Dec 22, 2019 1:57 am**

NI2M wrote:Hi,

I know that some of you are using Digital Audio Workstations or VST Hosts for audio processing which prompted these questions:

I use Voicemeeter Potato as a patch panel, and Reaper as my DAW. All of my audio, receive and transmit, goes through a Presonus Studio 192 Mobile interface.

1). I assume that you have a "dedicated" PC to run both the SDR and DAW or VST Host with plug-ins, correct?

Not really. I use my PC for everything, not just running the radio and the audio stuff.

2). If the answer to the above is in the affirmative, have you noticed audio drop-outs or other such "glitches" in your transmissions that would be traceable to using DAW or VST Host for your audio processing?

Affirmative or not, the answer is "no".

3). There are some excellent tutorials on how to build a PC that is truly dedicated to audio processing when employing a DAW or VST Host which have done away with a lot of the "bloatware" PCs come with that are deleterious on DAW and VST Host processing of audio. Depending on the extent to which one uses DAW (and especially the number and type of plug-ins) for audio processing a truly dedicated PC to audio may not be necessary, right?

This is a big "maybe". The sort of audio processing necessary for use with PowerSDR mRX or Thetis is substantially less than for someone doing multi-track recording or production. If you are running a reasonably powerful machine, say an i7 of any type, no special preparation is required. That said, you do need to check for gross problems using such tools as DPC Latency Checker and LatencyMon. If those tools indicate a problem, usually driver related, then the drivers and OS must be groomed to remove those problems first. Normally this just means using the right drivers.

4). Doesn't ANAN SDR in essence also process audio in the digital domain and you may also experience audio drop-outs, correct?

This is ABSOLUTELY correct, and, more importantly, the software client also processes the IF data stream, which normally has a relatively high sample rate, as high as 1.5MHz using Thetis. The IF data stream processing is actually much more demanding than the audio processing and is generally where problems are experienced.

The hardware in ANAN SDR with its FPGAs may make audio processing by the SDR rig inherently more efficient as well and audio drop-outs may not really be a concern at all, correct?

This is NOT true. The openHPSDR architecture is a "thick client" architecture. An IF data stream comes from the hardware (somewhat erroneously called a "radio", which is a misnomer since it takes both the PC and hardware unit to form a complete radio) and all audio, baseband and IF processing is performed via the software on the PC on both receive and transmit. The FPGA only performs the RF to IF and IF to RF conversions, as well as some general housekeeping.

5). Am I correct in thinking that even if one has a workstation grade machine with - say - an i9 CPU in it one might experience audio drop-outs as there are many functions such as automatic software updates and a myriad notifications that are going on in the "background" that potentially cause audio drop-outs, correct?

No, this is not normal at all with respect to DAW processing. However, the weak link in the chain is actually the IF DSP processing in either PowerSDR mRX or Thetis. If threads are blocked then real-time processing is interrupted and audible glitches will be heard. The SDR client software is working hard. The DAW software is not because you are only processing one or two channels.

6). Are there functions in a very hight performance PC when running just SDR that you have had to disable completely to not have any sort of audio drop-outs or other types of "glitches" when NOT using DAW or VST Host for audio processing and using what ANAN enables by itself?

For a reasonably powerful PC, the answer is "no". I'm running an i7-7700k and have no problems of this type. I take no special precautions other than to make sure all drivers are up to date and to select appropriately stable audio buffer sizes.

73,

Scott

Posted: **Sun Dec 22, 2019 5:47 pm**

Scott,

I appreciated your quick and to-the-point responses and comments. Below, interposed within the text, following "JJ", see my counters in italics.

73 de Juha
- NI2M
********************************************************************************************************************
I know that some of you are using Digital Audio Workstations or VST Hosts for audio processing which prompted these questions:

I use Voicemeeter Potato as a patch panel, and Reaper as my DAW. All of my audio, receive and transmit, goes through a Presonus Studio 192 Mobile interface.
JJ: "Potato" is an unknown entity for me, Will look that up.
JJ: I have a Steinberg UR22 (192 kHz USB/Audio interface) which works well for me. Will need to check to see if I could plug the UR22 into a faster USB port though (if my lap-top has one).

1). I assume that you have a "dedicated" PC to run both the SDR and DAW or VST Host with plug-ins, correct?

Not really. I use my PC for everything, not just running the radio and the audio stuff.

2). If the answer to the above is in the affirmative, have you noticed audio drop-outs or other such "glitches" in your transmissions that would be traceable to using DAW or VST Host for your audio processing?

Affirmative or not, the answer is "no".
JJ: I know from the designer of audio processing software that most of the latency comes from algorithms implemented in the plug-ins not the VST host or DAW.

JJ: As I mentioned, my machine is a simple i5 dual-core lap-top (I forget what the RAM and Cache sizes are now) so I think getting a desk-top with an i7 may be in order for "insurance." NOTE, some of the plug-ins (64 bit) I have in my chain exhibit a latency of 40 ms! A couple of limiters (or maximizers) I like having in the chain have a latency of around 20 ms so those 3 alone amount to a little less than 0.1 seconds! I also have compressors and EQs in the chain, but they exhibit very very low latency. Of course, this latency figure would be much lower if I had a multi-core i7 machine.

3). There are some excellent tutorials on how to build a PC that is truly dedicated to audio processing when employing a DAW or VST Host which have done away with a lot of the "bloatware" PCs come with that are deleterious on DAW and VST Host processing of audio. Depending on the extent to which one uses DAW (and especially the number and type of plug-ins) for audio processing a truly dedicated PC to audio may not be necessary, right?

This is a big "maybe". The sort of audio processing necessary for use with PowerSDR mRX or Thetis is substantially less than for someone doing multi-track recording or production. If you are running a reasonably powerful machine, say an i7 of any type, no special preparation is required. That said, you do need to check for gross problems using such tools as DPC Latency Checker and LatencyMon. If those tools indicate a problem, usually driver related, then the drivers and OS must be groomed to remove those problems first. Normally this just means using the right drivers.
JJ: I have used those tools for latency check-up and it seems that my machine is "border-line" and I think it better to get a real desk-top as I have plenty of lap-tops I can use for non-ham radio related tasks.
JJ: With what you are writing in mind I take it that whether the WiFi radio is ON or OFF doesn't make any difference? Is it actually better to leave the WiFi ON rather than switch if OFF? My thinking is that if the WiFi is turned OFF the PC may be wanting to "poll" for updates etc. even more often and with the WiFi OFF the PC may repeats those "polls" even more often? This is likely incorrect?

4). Doesn't ANAN SDR in essence also process audio in the digital domain and you may also experience audio drop-outs, correct?

This is ABSOLUTELY correct, and, more importantly, the software client also processes the IF data stream, which normally has a relatively high sample rate, as high as 1.5MHz using Thetis. The IF data stream processing is actually much more demanding than the audio processing and is generally where problems are experienced.
JJ: That makes sense.

The hardware in ANAN SDR with its FPGAs may make audio processing by the SDR rig inherently more efficient as well and audio drop-outs may not really be a concern at all, correct?

This is NOT true. The openHPSDR architecture is a "thick client" architecture. An IF data stream comes from the hardware (somewhat erroneously called a "radio", which is a misnomer since it takes both the PC and hardware unit to form a complete radio) and all audio, baseband and IF processing is performed via the software on the PC on both receive and transmit. The FPGA only performs the RF to IF and IF to RF conversions, as well as some general housekeeping.
JJ: Does the above suggest that it is impossible to MONITOR one's on-air transmission ( = MONITORING in the RF domain) as there is noticeable delay from start of speech to audible on-air speech caused by system latency? To me this isn't an issue as I can use the "CUE" feature in the VST Host to play audio clips (with raw audio from different mics) in a loop when tweaking parameters of the various plug-ins. I listen to myself (my transmit goes into a dummy of course) through a "flat" receiver to get a fairly good idea or "representation" of what I sound like in the RF domain. There IS a different between monitoring in the AF vs. RF domains; at least it is fairly clear in my own ears.

5). Am I correct in thinking that even if one has a workstation grade machine with - say - an i9 CPU in it one might experience audio drop-outs as there are many functions such as automatic software updates and a myriad notifications that are going on in the "background" that potentially cause audio drop-outs, correct?

No, this is not normal at all with respect to DAW processing. However, the weak link in the chain is actually the IF DSP processing in either PowerSDR mRX or Thetis. If threads are blocked then real-time processing is interrupted and audible glitches will be heard. The SDR client software is working hard. The DAW software is not because you are only processing one or two channels.

6). Are there functions in a very hight performance PC when running just SDR that you have had to disable completely to not have any sort of audio drop-outs or other types of "glitches" when NOT using DAW or VST Host for audio processing and using what ANAN enables by itself?

For a reasonably powerful PC, the answer is "no". I'm running an i7-7700k and have no problems of this type. I take no special precautions other than to make sure all drivers are up to date and to select appropriately stable audio buffer sizes.
JJ: What is Your main reasoning for using a DAW for "additional" audio processing? ANANs sound pretty good as they are when using a "flat" and wide enough a receive bandwidth? Just curious. To me, I like to experiment and learn and do like the additional controls that DAW/plug-ins afford.

73,

Scott

Posted: **Sun Dec 22, 2019 10:12 pm**

Juha,

https://www.vb-audio.com/Voicemeeter/potato.htm

Fabulous software. It is a true, ASIO compatible, virtual mixing board, with inserts to allow patching each channel though a DAW or VST host.

I have to agree that there is never a PC fast enough or powerful enough, and it is better to have it and not need it then need it and not have it.

However, and this is important: a faster PC will NOT reduce your latency for audio processing steps. For example, consider a digital filter. It will consist of some number of processing elements, and data is passed from one to the next at the audio sample rate, no faster, no slower. So if you have a plug-in that uses a 1024 point FFT, and your sample rate is 48KHz, it will take the FFT 1024/48000 seconds, or 21.3ms, to process that data. It cannot go any faster, the latency or delay is set by the sample rate and the number of steps in the DSP calculation that is being performed. It doesn't matter if you have a 5GHz processor, it cannot go any faster than it can obtain the necessary data.

What a faster PC can help with is insuring that your audio (and RF) processing threads are not blocked, are scheduled to run "on time", and in fact do run on time. But it can't help with latency here either, because the threads run in accordance with the sample rate and no faster.

The only way a faster PC can help latency is that it will often allow smaller interface buffer sizes. Thus, if you are not performing DSP functions on the data, the audio processing channel can have very low latency. With a good ASIO interface and a good PC, buffer sizes of 128 are usually easy to achieve. Again, the delay in seconds through each instance of such a buffer will be buffer size divided by sample rate in Hz.

WiFi drivers can be a source of DPC latency. Whether or not you have a problem depends on the specific WiFi NIC and it's drivers. You may or may not be able to have the WiFi activated when processing near-real-time audio.

With respect to monitoring: the MON function in the available software is not what you think it is. It provides a tap just before the data goes to the DAC. This means you have the worst case latency possible, and it means that if you turn on PureSignal the sound also includes the pre-distortion. If you are using MON to adjust your audio, leave PureSignal off while doing so. Also, as you alluded to in your previous post, it is best to use a recording of your unprocessed voice so that you do not have the psycho-acoustic problem of trying to listen to yourself talk with a long delay on the MON output.

Now for the million dollar question: why do I use a DAW? The reason it that I like the flexibility of it as a virtual mixing board. I could just use Voicemeeter Potato by itself, but instead I just use Potato as a glorified patch panel between Thetis, the DAW, and other software like WSJT-X and OBS. By doing it this way I can more easily create more complex mixing paths. I don't do any processing in the DAW, just mixing.

Now for the two million dollar question

: why don't I do any processing in the DAW, or Voicemeeter Potato for that matter? The reason is, Warren and I worked hard at putting pretty much everything that almost anyone would need for voice processing into both PowerSDR and Thetis. You've got mic gain, leveler, semi-parametric 10-band EQ, semi-parametric 10-band "CFC" compressor, semi-parametric 10-band post-EQ, and the look-ahead, soft-limiting ALC. That's two stages of EQ and three stages of compression (leveler, CFC compressor and ALC). Based on a lot of experience with the ESSB crowd, that covers about 95% of everyone's wants and desires. And in Thetis you also have a true look-ahead VOX and gate/expander, fully adjustable in every particular including a side-channel gate filter!

Take a look at the "how to set up your audio" topics in the PowerSDR sub-forum. I'm sure you'll agree that unless you are doing some really crazy stuff, the voice processing that is built into PowerSDR and Thetis is equal to, or in most cases superior to, what most folks are doing with their DAW or "rack" gear. Along with PureSignal, it allows the best sounding audio on the air, bar none. No DAW plug-in's required!

Finally, the Steinberg is a USB 2.0 device. Thus buffer sizes may have to be a bit larger, and driver latency is relatively long. I made the (very expensive) leap to a USB 3.0 interface, and this did help my latency quite a bit.

73,

Scott

Posted: **Tue Dec 24, 2019 1:10 am**

w-u-2-o wrote:Juha,

https://www.vb-audio.com/Voicemeeter/potato.htm

Fabulous software. It is a true, ASIO compatible, virtual mixing board, with inserts to allow patching each channel though a DAW or VST host.
JJ: Will take a look at this. Thanks Scott.

I have to agree that there is never a PC fast enough or powerful enough, and it is better to have it and not need it then need it and not have it.

However, and this is important: a faster PC will NOT reduce your latency for audio processing steps. For example, consider a digital filter. It will consist of some number of processing elements, and data is passed from one to the next at the audio sample rate, no faster, no slower. So if you have a plug-in that uses a 1024 point FFT, and your sample rate is 48KHz, it will take the FFT 1024/48000 seconds, or 21.3ms, to process that data. It cannot go any faster, the latency or delay is set by the sample rate and the number of steps in the DSP calculation that is being performed. It doesn't matter if you have a 5GHz processor, it cannot go any faster than it can obtain the necessary data.
JJ: This is basic physics really and I appreciate it that you made this point. A good reminder indeed.

What a faster PC can help with is insuring that your audio (and RF) processing threads are not blocked, are scheduled to run "on time", and in fact do run on time. But it can't help with latency here either, because the threads run in accordance with the sample rate and no faster.

The only way a faster PC can help latency is that it will often allow smaller interface buffer sizes. Thus, if you are not performing DSP functions on the data, the audio processing channel can have very low latency. With a good ASIO interface and a good PC, buffer sizes of 128 are usually easy to achieve. Again, the delay in seconds through each instance of such a buffer will be buffer size divided by sample rate in Hz.

WiFi drivers can be a source of DPC latency. Whether or not you have a problem depends on the specific WiFi NIC and it's drivers. You may or may not be able to have the WiFi activated when processing near-real-time audio.
JJ: I switch my WiFi off and I can see the CPU load decrease somewhat. If I didn't consistently observe any load reduction I wouldn't bother messing with the WiFi radio in my PC.

With respect to monitoring: the MON function in the available software is not what you think it is. It provides a tap just before the data goes to the DAC. This means you have the worst case latency possible, and it means that if you turn on PureSignal the sound also includes the pre-distortion. If you are using MON to adjust your audio, leave PureSignal off while doing so. Also, as you alluded to in your previous post, it is best to use a recording of your unprocessed voice so that you do not have the psycho-acoustic problem of trying to listen to yourself talk with a long delay on the MON output.
JJ Thanks for making the clarification in regards to the MON function.

Now for the million dollar question: why do I use a DAW? The reason it that I like the flexibility of it as a virtual mixing board. I could just use Voicemeeter Potato by itself, but instead I just use Potato as a glorified patch panel between Thetis, the DAW, and other software like WSJT-X and OBS. By doing it this way I can more easily create more complex mixing paths. I don't do any processing in the DAW, just mixing.
JJ: This is a great tip Scott.

Now for the two million dollar question : why don't I do any processing in the DAW, or Voicemeeter Potato for that matter? The reason is, Warren and I worked hard at putting pretty much everything that almost anyone would need for voice processing into both PowerSDR and Thetis. You've got mic gain, leveler, semi-parametric 10-band EQ, semi-parametric 10-band "CFC" compressor, semi-parametric 10-band post-EQ, and the look-ahead, soft-limiting ALC. That's two stages of EQ and three stages of compression (leveler, CFC compressor and ALC). Based on a lot of experience with the ESSB crowd, that covers about 95% of everyone's wants and desires. And in Thetis you also have a true look-ahead VOX and gate/expander, fully adjustable in every particular including a side-channel gate filter!
JJ: I am not sure if I am in the 5% of the "ESSB crowd," but what you have done is great! Some of my audio processing may fly in the face of distortion-free audio as I do use tape emulation plug-ins to introduce "positive" distortion. As you may know well, a "tape machine" is somewhat similar to having a DEQ in the chain AND also a saturator. I am not trying to be "cute" or anything but I actually hear the difference when I drive my audio hard into a "tape machine" emulation plug-in and I like what I am hearing. Also, I am experimenting with an EQ that tracks different frequencies (I can set the bands myself) of my voice in real time. Very cool from an experimenters perspective as I can create some "unique voice signatures" with my fundamental for instance. I can have boost/cuts (including different Q values) at the fundamental (around 100 Hz in my case) frequency and add boosts/cuts at 2rd, 3rd,4th etc. order frequencies while all of these (AND the cuts or boost that I have specified) are shifting around or following the even/odd frequencies in real-time as I speak. So, I can add some additional "growl" or "bite" into my fundamental or do it by having the EQ boost the 3rd fundamental, many permutations exist. Yes, I realize some of this "fancy foot-work" may not translate to ham radio at all, but I do enjoy these sorts of tweaks and sometime I stumble upon some pretty nice sounding audio as listened to through a flat RX in the RF domain.

Take a look at the "how to set up your audio" topics in the PowerSDR sub-forum. I'm sure you'll agree that unless you are doing some really crazy stuff, the voice processing that is built into PowerSDR and Thetis is equal to, or in most cases superior to, what most folks are doing with their DAW or "rack" gear. Along with PureSignal, it allows the best sounding audio on the air, bar none. No DAW plug-in's required!
JJ: I guess I may be one of those fellows doing some "really crazy stuff."

Finally, the Steinberg is a USB 2.0 device. Thus buffer sizes may have to be a bit larger, and driver latency is relatively long. I made the (very expensive) leap to a USB 3.0 interface, and this did help my latency quite a bit.
JJ: That is right! My garden-variety lap-top doesn't have USB 3.0. The PreSonus Audio/USB interface you are using is that "(very expensive) leap" I assume. I have spent that kind of money and then some on plug-ins alone so far.........

JJ: Does your PC have a SSD? Again, that alone won't really help much concerning latency the way my in-the-box audio processing environment is configured as you rightly pointed out. However, having a SSD instead of a hard disk (hard disks are still very inexpensive and constitute a good storage medium though) will at least enable the software to run faster. There again a SSD would me an exercise in futility unless my system overall is configured in a way that could actually benefit from having a SSD instead of a hard disk.

73 de Juha
NI2M

73,

Scott

Posted: **Tue Dec 24, 2019 2:07 am**

The only thing that still spins around here are the 2TB drives in my NAS.

Posted: **Tue Dec 24, 2019 4:31 am**

Apache Labs Community Forum

Audio drop-outs when using DAW or VST Host

Audio drop-outs when using DAW or VST Host

Re: Audio drop-outs when using DAW or VST Host

Re: Audio drop-outs when using DAW or VST Host

Re: Audio drop-outs when using DAW or VST Host

Re: Audio drop-outs when using DAW or VST Host

Re: Audio drop-outs when using DAW or VST Host

Re: Audio drop-outs when using DAW or VST Host