Beta Test: Multitone Loopback Analyzer software

[Sokel]

So I'm assuming classical test tones are based on 432 hz?

Or 440Hz but that's another debate coming as far as Bach's era

 

[Sokel]

If it helps,increasing the buffer by 10ms each time flashes increase too with it at an analogous frequency.
They even have rhythm!

(Good news is that result is ok most of the times and more good news is that EMU works fine with 40ms where the flashes stop )

 

[pkane]

If it helps,increasing the buffer by 10ms each time flashes increase too with it at an analogous frequency.
They even have rhythm!

(Good news is that result is ok most of the times and more good news is that EMU works fine with 40ms where the flashes stop )

Buffer size still matters, and while the latest versions improve on latency, you may still need to increase buffer size on slower computers and/or larger FFT sizes. Without rewriting the audio library, I'm not sure I can speed up the processing much further. I'll try, but it's unlikely to make a large difference. I might consider using a dedicated CPU core for spectrum processing/real-time display and use other CPU cores for playing and capturing sound. I'm not sure it'll make a big difference, but might be worth a try.

 

[Sokel]

Buffer size still matters, and while the latest versions improve on latency, you may still need to increase buffer size on slower computers and/or larger FFT sizes. Without rewriting the audio library, I'm not sure I can speed up the processing much further. I'll try, but it's unlikely to make a large difference. I might consider using a dedicated CPU core for spectrum processing/real-time display and use other CPU cores for playing and capturing sound. I'm not sure it'll make a big difference, but might be worth a try.

Yes,but that changed just two versions ago,I know because my standard 125ms setting (kind of a sweet point where I can do anything) creates these flashes now.
Before that I could use all buffer sizes available (2ms to 500ms) with no major problems (some just created some short of slow motion preview window) ,now I can only use 2ms to 40ms,after that preview window goes crazy (all thought most of the times the results are not,they are ok).

 

[pkane]

Yes,but that changed just two versions ago,I know because my standard 125ms setting (kind of a sweet point where I can do anything) creates these flashes now.
Before that I could use all buffer sizes available (2ms to 500ms) with no major problems (some just created some short of slow motion preview window) ,now I can only use 2ms to 40ms,after that preview window goes crazy (all thought most of the times the results are not,they are ok).

I don't think anything's changed with regards to latency since 1.0.90. But let me know if you can use lower buffer sizes with 1.0.90 than with 1.0.91 or later.

 

[Rantapossu]

I don't think anything's changed with regards to latency since 1.0.90. But let me know if you can use lower buffer sizes with 1.0.90 than with 1.0.91 or later.

I think that the change happened between V1.0.89 and V1.0.90 "second edition", where preview screen was rebuilded by @Sokel 's request. Do you still have the initial V1.0.90 "first edition" (without the preview screen rework) compiled in your archive? I didn't manage to download it, because it was online only for a couple of hours.

 

[Sokel]

The one I have marked as "buffer changed" is 1.0.89.0, 8/12/2023.
The one that caused the flashes is the very last fix.
I could use low latency before as well.

 

[pkane]

The one I have marked as "buffer changed" is 1.0.89.0, 8/12/2023.
The one that caused the flashes is the very last fix.
I could use low latency before as well.

I'm now fully confused Let's see if we can figure this out by answering a few questions:

1. The problem occurs only with the preview box, yes?
2. It manifests itself as jumps in the spectrum display in the preview box?
3. The problem started happening after 1.0.89?
4. The fix that I added to clear the buffer (very last fix) made the preview box flash more than before during the recording? That makes very little sense to me, as the change was very minor and only applies to after the recording is complete. I can't see how it can cause flashing in the middle of the recording.
5. Or am I totally misunderstanding what you are reporting?

(@Sokel , @Rantapossu please respond to these individually, as I'm not even sure that what you both describe is the same or different issue)

Regards,

-Paul

 

[Sokel]

1. Yes (mostly,going very high,more than 150ms affects measurement too)
2. Yes
3. No,it was ok until the very last fix,it then appeared.
4. Yes,but not in the middle,right from the start.
5. no,you didn't misunderstood

 

[Rantapossu]

I'm now fully confused Let's see if we can figure this out by answering a few questions:

1. The problem occurs only with the preview box, yes?
2. It manifests itself as jumps in the spectrum display in the preview box?
3. The problem started happening after 1.0.89?
4. The fix that I added to clear the buffer (very last fix) made the preview box flash more than before during the recording? That makes very little sense to me, as the change was very minor and only applies to after the recording is complete. I can't see how it can cause flashing in the middle of the recording.
5. Or am I totally misunderstanding what you are reporting?

(@Sokel , @Rantapossu please respond to these individually, as I'm not even sure that what you both describe is the same or different issue)

Regards,

-Paul

I have two problems (At least with Multitone)

The first one is the buffer looping at the end. Latest build fixed it with my Audient iD24, but it didn't fix Asus Xonar U7 and E-MU 0202 (@Sokel uses it's bigger brother 0204). Should be fixable with adding digital zero to end of the source file, so that the buffer is empty when looping happens. Worked with V1.0.89, is broken at least with V1.0.90 "second edition" and is partly fixed with 1.0.92 latest preview build. I don't have V1.0.90 "first editition" to verify.

And then the second. I think that @Sokel means this:

This flashes between the "normal" view and the above view about 1-2 times a second with E-MU 0202 (Maybe with other cards too, I haven't tried). Doesn't happen with smaller buffers (The above is taken with 125 ms buffer).

1. Yes
2. Yes
3. Yes. I don't know exactly when. I don't have all the versions available and I started to study this only recently.
4. This happens all the way from start to end.
5. No

 

[Sokel]

I also monitored the demands with my ancient machine:
For 1Khz,256 FFT,8 averages,96Khz sample rate CPU goes from 4.5% to 13.5% and memory demands are from 250 to 600 MB leaving overall memory another 4 GB available as it reports.

 

[Rantapossu]

There were several preview versions of V1.0.89, but I have a hunch that the "All ASIO channels -enabler" for @Grooved might have done something to ASIO support (I use ASIO and so does @Sokel). The looping doesn't happen with V1.0.89 (Aug 12 11:00PM build). Then there is large gap of versions and the next versions that I have are from September and all of them loops, except the very latest test build, and only with my Audient. E-MU and Xonar loops still.)

 

[pkane]

1. Yes (mostly,going very high,more than 150ms affects measurement too)
2. Yes
3. No,it was ok until the very last fix,it then appeared.
4. Yes,but not in the middle,right from the start.
5. no,you didn't misunderstood

Ok, thanks! I understand the issue, and I think it was present before, just not all the time. I'll see if there's something that's changed, although I can't imagine that last fix (for clearing ASIO buffers at the end of the signal) could possibly affect the start

 

[pkane]

I have two problems (At least with Multitone)

You don't really want my help with any problems not related to Multitone, believe me!

The first one is the buffer looping at the end. Latest build fixed it with my Audient iD24, but it didn't fix Asus Xonar U7 and E-MU 0202 (@Sokel uses it's bigger brother 0204). Should be fixable with adding digital zero to end of the source file, so that the buffer is empty when looping happens. Worked with V1.0.89, is broken at least with V1.0.90 "second edition" and is partly fixed with 1.0.92 latest preview build. I don't have V1.0.90 "first editition" to verify.

Can you please capture and post the waveform from Xonar U7 or EMU 0202 that exhibits this behavior and post the waveform plot (or Audacity, as you did before)? I'm trying to understand how there can be anything at the end if the buffers are completely zeroed out by my last change.

And then the second. I think that @Sokel means this:

View attachment 314646
This flashes between the "normal" view and the above view about 1-2 times a second with E-MU 0202 (Maybe with other cards too, I haven't tried). Doesn't happen with smaller buffers (The above is taken with 125 ms buffer).

So a larger buffer causes more jumps? That's is exactly the opposite of everything I've experienced... strange. Can you try increasing the buffer to a maximum possible value and see if the jumps still occur in the middle of a recording?

1. Yes
2. Yes
3. Yes. I don't know exactly when. I don't have all the versions available and I started to study this only recently.
4. This happens all the way from start to end.
5. No

Thank you!

 

[Sokel]

So a larger buffer causes more jumps? That's is exactly the opposite of everything I've experienced... strange. Can you try increasing the buffer to a maximum possible value and see if the jumps still occur in the middle of a recording?

Yes,that's the strange thing,the bigger - the worst!
At the largest (500ms) it does slow motion (more like interrupted)+ jumps for all duration.
But the result is solid,does not affect it,same THD+N as always.

 

[Sokel]

Also Deltawave is showing it somehow strange:





40 ms




500 ms

There's some kind of intermission in the middle to the end.

 

[Rantapossu]

You don't really want my help with any problems not related to Multitone, believe me!

Can you please capture and post the waveform from Xonar U7 or EMU 0202 that exhibits this behavior and post the waveform plot (or Audacity, as you did before)? I'm trying to understand how there can be anything at the end if the buffers are completely zeroed out by my last change.

Sure:

The whole measurement. Red part should be silent:

Let's take one peak...

and zoom:

6615 samples between the spikes:

Buffer is 150 ms above. So 0.15 * 44100 samples/s = 6615 samples -> Matches

So it repeast the last 6615 samples all over again...

So a larger buffer causes more jumps? That's is exactly the opposite of everything I've experienced... strange. Can you try increasing the buffer to a maximum possible value and see if the jumps still occur in the middle of a recording?

Jumps doesn't happen, if the buffer is below 125 ms. Jumps happen 125 ms to 500 ms (Max). The bigger the buffer the slower the jumps. More sluggish, I say...

And all the way from start to end...

 

[pkane]

Sure:

The whole measurement. Red part should be silent:

View attachment 314667


Let's take one peak...

View attachment 314668



and zoom:

View attachment 314670

6615 samples between the spikes:


View attachment 314671

Buffer is 150 ms above. So 0.15 * 44100 samples/s = 6615 samples -> Matches

So it repeast the last 6615 samples all over again...



Jumps doesn't happen, if the buffer is below 125 ms. Jumps happen 125 ms to 500 ms (Max). The bigger the buffer the slower the jumps. More sluggish, I say...

And all the way from start to end...

Are you sure that’s the latest version? Hard to believe that it is. Can you try to reinstall?

 

[Rantapossu]

Are you sure that’s the latest version? Hard to believe that it is. Can you try to reinstall?

I'm positive, but I can reinstall if needed. Tested with two computers (Win 10 and Win 11) and with both Xonar and E-MU. Same version works with Audient. Maybe the new Audient obey your cache cleaning command and the ancient cards don't.

 

[Rantapossu]

Are you sure that’s the latest version? Hard to believe that it is. Can you try to reinstall?

I uninstalled MT first...

Taken again from this post (Re-downloaded a second ago):

I think I understand... although still can't reproduce. Can you please try this version?

https://app.box.com/s/ue7ll9xmvwogst817x2l1xg09opvgy47

Sep 23 10.20PM build...

And the result: