A "jangling" artifact on TX and RX audio, related to filter size?

User avatar
kc2rgw
Posts: 165
Joined: Mon Jun 22, 2020 5:44 pm

A "jangling" artifact on TX and RX audio, related to filter size?

Postby kc2rgw » Tue Apr 13, 2021 3:44 pm

I thought I'd asked this before but maybe not. I've noticed an odd "jangling" like shaking a bag of metallic objects or shaking a metal chain that is an artifact on voice peaks that I can hear on receive with strong signals and it can be heard on my TX audio as well when signals are strong for the receiver.

It seems that it comes in if I run my buffer size at 64 and my filter size at 8192 but smooths out or goes away when I run 128 buffer and 4096 for the filter size.

Just curious what the cause is if this makes sense to someone else and if what I'm noticing makes sense with the parameter changes?

Here's a screen shot of the DSP->Options screen when I don't hear it.

I've noticed this whether feeding audio via VAC or via the "analog" path through the line input. This is a 7000DLE MKII with Thetis and 2.0.0b firmware I believe but has been something I've noticed with multiple combinations of software and firmware.

setup-dsp-options.png
setup-dsp-options.png (49.29 KiB) Viewed 3928 times
SA3ATF
Posts: 95
Joined: Mon Apr 10, 2017 9:42 pm
Location: Vasterasen, Bispgarden, Sweden JP82HX
Contact:

Re: A "jangling" artifact on TX and RX audio, related to filter size?

Postby SA3ATF » Tue Apr 13, 2021 5:12 pm

Nice to read there are more who can hear this phenomenon.
I pointed this out a few years ago, but no one reacted.
I don’t suffer from insanity, I enjoy every second of it!
User avatar
w-u-2-o
Posts: 5540
Joined: Fri Mar 10, 2017 1:47 pm

Re: A "jangling" artifact on TX and RX audio, related to filter size?

Postby w-u-2-o » Tue Apr 13, 2021 6:51 pm

Buffer size should have nothing to do with this issue. It would be good if you would confirm that is the case.

What this is really about is the transient response of the filter skirts. This is not ringing per se, but it is a time domain, transient response issue.

There are three things that cause transient response effects to increase or decrease for those people who can hear them. They are: filter size, which defines the skirt slope, filter type, which defines the basic transient response characteristics, and filter window, which also affects skirt slope.

Even at the smallest filter length setting, 1024, the slope of the skirts of the passband filters are generally much steeper than that found on any radio other than the Flex. It is rarely necessary to need steeper skirts than the minimum setting. Steeper slopes generally equate to more audible artifacts.

The low latency filter type, which is really a minimum phase FIR, as opposed to the linear phase FIR filter type, tends to cause those who are more sensitive to these transient response characteristics to notice the issue at smaller filter sizes. Thus, if you are willing to give up latency, switching to the linear phase filter type can help.

Another thing that can help is changing to a BH-7 Filter Window setting. The BH-7 mode will provide deeper stop band rejection at the expense of filter skirt slope. Thus using the BH-7 setting can help mitigate the issue somewhat. It is easy to have your cake and eat it too with BH-7 if you still want the steep skirts. Just double the length you were using with BH-4 and you will approximate what you had previously with BH-4.

The softest, most linear transient response can therefore be achieved using a 1024 length, BH-7 window, a Linear Phase filter type.

As for myself, I can't hear this issue at all! I happily run Low Latency filters of length 4096 with the BH-7 window (equivalent to a 2048 length with a BH-4 window).
User avatar
kc2rgw
Posts: 165
Joined: Mon Jun 22, 2020 5:44 pm

Re: A "jangling" artifact on TX and RX audio, related to filter size?

Postby kc2rgw » Wed Apr 14, 2021 1:59 am

Thanks Scott, I was kind of grabbing params until I didn’t hear it anymore. I’ll check tomorrow by putting the buffer size back to 64.

I was initially at buffer 64 , size 8192, low latency and the 7 filter when someone pointed it out on my TX audio...and I had been noticing it in my recordings while setting my audio up.

I basically changed the size as shown in the screen shot and it either went away or lowered enough so as not to be very prominent. I’ll see if the buffer size makes any audible effect I can notice.
User avatar
rbduck
Posts: 327
Joined: Tue Dec 03, 2019 1:49 pm

Re: A "jangling" artifact on TX and RX audio, related to filter size?

Postby rbduck » Wed Apr 14, 2021 8:57 am

I'm just curious. Are you hearing this with NR or NR2 turned on? I was trying to duplicate the condition, The only jangling I'm hearing is with NR2 turned on. Otherwise, I don't hear a thing.


kc2rgw wrote:I thought I'd asked this before but maybe not. I've noticed an odd "jangling" like shaking a bag of metallic objects or shaking a metal chain that is an artifact on voice peaks that I can hear on receive with strong signals and it can be heard on my TX audio as well when signals are strong for the receiver.

It seems that it comes in if I run my buffer size at 64 and my filter size at 8192 but smooths out or goes away when I run 128 buffer and 4096 for the filter size.

Just curious what the cause is if this makes sense to someone else and if what I'm noticing makes sense with the parameter changes?

Here's a screen shot of the DSP->Options screen when I don't hear it.

I've noticed this whether feeding audio via VAC or via the "analog" path through the line input. This is a 7000DLE MKII with Thetis and 2.0.0b firmware I believe but has been something I've noticed with multiple combinations of software and firmware.

setup-dsp-options.png
73
Ruben
NB4R
Apache-Labs Anan 7000DLE MKII Black -- Thetis 2.10.3.6 dev_2 -- Windows 11
User avatar
kc2rgw
Posts: 165
Joined: Mon Jun 22, 2020 5:44 pm

Re: A "jangling" artifact on TX and RX audio, related to filter size?

Postby kc2rgw » Wed Apr 14, 2021 11:05 am

Yup with NR2, it’s almost always on. Good point.
K1LSB
Posts: 639
Joined: Wed Feb 05, 2020 5:25 pm

Re: A "jangling" artifact on TX and RX audio, related to filter size?

Postby K1LSB » Wed Apr 14, 2021 12:43 pm

NR (and NR2) has no effect on TX audio. The jangling is on your TX audio too so turning off NR2 won't help that.

Mark
User avatar
kc2rgw
Posts: 165
Joined: Mon Jun 22, 2020 5:44 pm

Re: A "jangling" artifact on TX and RX audio, related to filter size?

Postby kc2rgw » Wed Apr 14, 2021 1:34 pm

Yup, just tested it this morning. NR2 has no effect. Buffer size didn't change it either. The filter size is where it was with the combo of the low-latency option. At 4096 it's so minimized, nearly gone, 8192 it becomes more prevalent again. So easy enough just run it at 4096 for me and it works well enough.

Was just trying to figure out where/why/how it happens and honestly most people don't hear it on TX it's just people who are very discerning about audio who have noticed it. I find it very obvious but a lot of people don't seem to hear it.
User avatar
rbduck
Posts: 327
Joined: Tue Dec 03, 2019 1:49 pm

Re: A "jangling" artifact on TX and RX audio, related to filter size?

Postby rbduck » Wed Apr 14, 2021 2:02 pm

I was referring to the receive audio and not the TX audio. I've never encountered that on TX audio.

kc2rgw wrote:Yup with NR2, it’s almost always on. Good point.
73
Ruben
NB4R
Apache-Labs Anan 7000DLE MKII Black -- Thetis 2.10.3.6 dev_2 -- Windows 11
User avatar
w-u-2-o
Posts: 5540
Joined: Fri Mar 10, 2017 1:47 pm

Re: A "jangling" artifact on TX and RX audio, related to filter size?

Postby w-u-2-o » Wed Apr 14, 2021 2:04 pm

The other thing to consider is that this setting is much less important on transmit. As mentioned above, at 1024 the skirts are already much steeper than any other radio out there save the Flex. You are not going to look or sound bad at 1024 transmit. So you can knock that setting down to minimum on transmit while still having brick wall filtering on receive if you like, that's why the settings are separate for TX and RX.

Return to “Thetis”