Decent/Cheap/Slower USB audio that works?

[mnopneal]

Ok, we have a Ubuntu server on a Business Comcast connection at church with Jacktrip and Jamulus running.

We have 7 home built Virtual Studio devices with HiFiBerry cards. One of them was intended for the sanctuary. The others are already deployed at musicians homes and are working.

We now have a microSD of Jambox dinked from these discussions to run with the HiFiBerry card. It appears to work.

HOWEVER... Those HIFI cards are expensive AND scarce.

This JamBox has near zero network latency to the server downstairs. All the other VS boxes will have 30+ms of network latency.

Could we not use one of the cheap but decent USB adapters on this, and use that HIFI card for another singer?

How about the UGreen adapter? Or is there another cheap one that works OK out of the box? This is fed from, and feeding a Behringer X32 rack mixer, so levels is not a problem.

regards,

Neal Rhodes

[kdoren]

@mnopneal I've been experimenting with some low-cost USB adapters. All of them "work" in the sense that I can get audio going in & out. Some are unusable due to bad latency (i.e. Plextone GS3), or power supply noise (i.e. Vantec NBA-200U).

Here are 3 that might do what you want. All have low latency and good headphone output. All have a single mono mic input capture channel.

• low-cost USB audio interfaces
  • TechRise ($12): https://www.amazon.com/gp/product/B01J7P0OGI
    Has a physical volume control,
  • UGreen ($14): https://www.amazon.com/gp/product/B01N905VOY
  • Sabrent ($7): https://www.amazon.com/gp/product/B00IRVQ0F8

To control internal settings, use the "pimixer" program. I'll put this on the desktop in the next Jambox release, but for now you can install with "sudo apt install pimixer", and launch from the Pi menu under "Preferences" -> "Audio Device Settings".

I tried this with a Pyle dynamic mic (through an adapter to 3.5mm). In all cases I had to set mic capture level to 100% or close, and I thought the hiss from the preamp was rather high,. If you can find a higher-outout mic, you could lower the capture level which would help.

The physical volume control on the TechRise is a nice feature. UGreen & Sabrent have software-switchable AGC, which I needed to set to "on" to get enough Mic Gain. I didn't hear any actual AGC at work, so it might just be a boost switch.

The single capture channel will go by default to Jamulus left channel, then Jamulus by default (mono up) will mix to both. You can get higher level (including higher noise) by sending the capture channel to both Jamulus input channels (see the topic "How do I change patches).

My thinking is: If you are only building 1 or a few boxes, get the Behringer UM2 for $49, it will work well with no risk and minimal hassle. If you are building a lot of boxes and price is a big issue, try some tests with one of the adapters above (I'd start with TechRise for the physical volume control). Try to find a mic with high enough output that you can lower the mic capture volume for lower noise.

Please report back here with your findings!

[mnopneal]

Thanks so much for the reply.

We did try the UGreen way back in October and were not impressed with the gain. We'll look at the TechRise next. We've got two on order for arrival Monday. Don't ya love Prime Shipping with Free Returns?

Since then, looking for the best/cheap solution for singer/musician couples recording from home, I tested a $5 desktop mic, and then one of our couples found this $15 alleged condenser mic: https://www.amazon.com/gp/product/B08GZSQNT7/ref=ppx_yo_dt_b_asin_title_o00_s01?ie=UTF8&psc=1 so I made a head-to-head test, using Audacity, on a notebook, with record input set to 80%. Tested at 3", 12", 12" at a 45 degree angle, 12" at a 90". The above mic has 15% one star reviews, but I gotta say, it was hotter in all those tests, and had more of the proximity effect for consonants than the desktop mic. Not as good as a real MXL 990 at 24", but workable for ensemble recording. (there's a real limit on how close two people and two instruments can get. Just make sure you get one with free returns, 'cause there seems to be a 15% DOA.

This mic is so much hotter at the same distance that it should make a difference. Also, it sounds good at 12" away. Sure we could give every singer an SM58, AND a XLR cable, AND a mic stand, AND... but $15 and we're done sounds good to me.

I also understand the extra latency over a UM2, and all our previous boxes were HiFiBerry, but it boils down to X units with Y amount of grant money covers X people.

I DO have to say I'm impressed and grateful that you turned on the Mic Bias Voltage on the HiFi card. THAT was a hassle with the Virtual Studio.

regards,

Neal

[mnopneal]

I got a pair of TechRise USB adapters in today. With the above mic, and turning up volume in alsamixer to 100%, I found just a vocal at 12" resulted in Top-of-the-Green volume. At 3" I hit top of the red easily.

Initially I was doubly thrilled to find two headphone jacks, but it just turns out that one of them is pure headset, the other headset+microphone. Still, good volume, low noise.

Jamulus reports 34ms Ping time, overall Delay 58ms. How is it measuring overall?

[kdoren]

@mnopneal Jamulus is measuring overall delay that it knows about, which is the path through the server, plus the buffer settings.

But Jamulus doesn't know about delays inside your audio device. For most of the USB devices I have tried, the delays are not bad, and in a narrow range. However there are exceptions which can be pretty bad (i.e. Plextone GS3).

2 ways to measure. This is mostly useful to compare differences between interfaces. Part/most of this measurement (the buffering) is included in Jamulus Overall delay. Before measuring, stop Jamulus, etc. Review settings in /etc/jackdrc.conf, then restart jack ("sudo systemctl restart jack")

• jack_delay
  • connect a loopback cable between audio outputs and inputs
  • Install and start jack_delay

    sudo apt install jack-delay
    jack_delay -I system:capture_1 -O system:playback_1
    

  If your loopback cable is correct (channels not swapped) it should show the delay.
• Audacity
  • you can generate a clicktrack in Audacity, then make an overdub recording while playing the clicktrack.
  • then you can zoom in and visually see the delay.
  • this method can work if loopback path is holding a mic up to headphone or speaker.
  • see this guide: https://manual.audacityteam.org/man/latency_test.html
  • also need to sey: Edit -> Preferences -> Devices -> JACK Audio -> Track shift after record -> "0"

FYI I didn't turn on mic bias for the hifiberry card, that might be a default setting or got set there another way.
Best way to get to internal settings is with pimixer: "sudo apt install pimixer", then launch with Pi menu -> preferences -> Audio Device Settings. Changes made here appear to be sticky (survive a reboot).

[mnopneal]

Thank you. I gotta say, that you have made PI audio what I'd call "Survivably Neat and Fun". After retiring from decades of IT consulting, working with Jambox is a breeze. Working with Virtual Studio, is work.

regards,

Neal

[mnopneal]

Ok, so let's see if I understand your math here. Jamulus reports ping time as 36ms, overall Delay as 56ms. So 20ms is buffering. jack_delay reports 26ms. Of which 20ms was already known as buffering. So, latency in USB audio is... around 6ms?

BTW, probably good to turn off Mic Bias Voltage before patching input to output. I noticed a tiny spark when connecting cable.

Cheers,

Neal

[kdoren]

I think that is approximately correct but I'm not sure how exact you can get with it.

If you are at 26ms for TechRise with jack_delay, you can probably reduce delay by reducing NPERIODS in /etc/jackdrc.conf.

What is your NPERIODS setting currently?

[mnopneal]

I just did an apples/apples comparison of volume of TechRise with HiFiBerry.

Same device playing music, two different PI boxes, same headset.

HiFiBerry: 40dB
TechRise: 60dB

[ve3meo]

That cheap microphone has a stated sensitivity of -55dB which is pretty good. It's not so cheap from Amazon Canada at CAD$55.

I don't recall if I reported on some cheap adapters but one is plugable(TM) USB-AUDIO which has more like a line level sensitivity on its inputs which are stereo but left and right are reversed. It's also a bit noisy but I was keen on the 2-channel mic/line input. https://www.amazon.com/Plugable-Headphone-Microphone-Aluminum-Compatible/dp/B00NMXY2MO/ref=sr_1_3?dchild=1&keywords=plugable+usb-audio+adapter&qid=1613679304&sr=8-3

[mnopneal]

So, it would seem the BIG ASK these days is whether the HiFiBerry is worth the dollars and the wait. Here is what I know:

Techrise: Jambox Statistics: Ping 38ms, Overall Delay 52ms. Jack_Delay 26ms.
HiFiBerry:Jambox Statistics: Ping 37ms, Overall Delay 52ms. Jack_Delay 13ms.

So, we surmise HiFiBerry is 13ms faster, in fact near zero? Which may be important in some contexts. OTOH, both UGreen AND Techrise are MUCH hotter mic preamps, and much louder headphone outputs. For some people HiFiBerry is unusable without a $30 headphone amp. Depending on where your group lives, and where they land in the network latency spectrum, rolling out a solution today with either one might be reasonable. Especially when we ask how much longer we're in this situation, and our expectations. I think we are sticking with a strategy of HiFiBerry on AT&T singers, and USB on Comcast singers, and the 12ms slower AT&T is mitigated by the 13ms faster HiFiBerry.