Forum Replies Created
8 November 2023 at 7:52 pm #732
I doubt you can load 12 & 24 kHz filters at all. If it could load these sample rates, keep in mind there’s a maximum tap length available for FIR so the 12 kHz filter – upsampled to 48 kHz – may get truncated, which could dramatically change the filter response.
The Lake FIR 3-way and FIR XP modules are the only modules I’m aware of that use those lower sample rates.
EA Support8 November 2023 at 6:48 pm #729
I don’t know. You’ll have to try it. The sample rate is included in the GKF file that we export for Dynacord, so IRIS-Net should be able to tell you if the sample rate is not supported.
Take a look at EV and Dyncord on
From the 2nd link: “Note: IRIS-Net appears to load both 48 kHz and 96 kHz FIR filter files, and convert them appropriately to the sample rate of the processor.”
If there’s information missing, let us know and we’ll add it.
EA Support7 November 2023 at 1:37 pm #727
What DSP is this? The file looks like a format we can implement to export from our products.
Here’s a workflow to get to the FIR data.
1. Copy the c0= ….. c383=”-1.497382E-07″ stuff to a text editor and use find/replace to remove all the ”
2. Copy the c0= ….. c383=-1.497382E-07 from the text editor into an empty worksheet in Excel
3. In Excel, select “Data” “Text to Columns” and use a space ” ” delimiter to break the single cell into 384 cells in a row.
4. Copy the 384 cells
5. On a vacant cell in Excel, right-click and select “Paste Special” “Transpose”. You’ll now have a column starting with c0=1.729377E-10 in the top cell.
6. Select the 384 cells (in the column) and select “Data” “Text to Columns” with an equals sign = as a delimiter. Now you have two columns; one starting with c0 and one with 1.73e-10.
7. Select all the number cells, right-click and select “Format” “Scientific” with 6 decimal places.
8. Copy the 384 cells back to a text editor and save as a TXT file. The results follow below.
9. Start “FIR Designer” and don’t load anything on the Import tab. On the Target tab and File sub-tab, press Load to load the TXT file. Uncheck Magnitude only and Phase only. Press “Find Peak”.
10. In “FIR Designer”, on the Target tab and Curve sub-tab, uncheck “Use”.
Now you have the filter loaded as a target and you can use the rest of the FIR Designer workflow to make a new FIR filter that either matches the loaded filter, or is slightly different to the loaded filter.
-1.497382E-074 October 2023 at 8:10 am #717
I can confirm we see a difference on Windows, but not on macOS.
In FIR Designer M version 4 we implemented significant speedups to many of the underlying computations. These have not yet been ported into FIR Designer but will be in FIR Designer version 4.
EA Support4 October 2023 at 6:13 am #714
Exactly which versions of FIR Designer and FIR Designer M are you comparing and on which OS (macOS or Windows)? Also, please email us a FIR Designer project and FIR Designer M system that shows the effect so that we can investigate quickly.
13 September 2023 at 8:27 pm #708
- This reply was modified 4 months, 3 weeks ago by EA Support.
Yes, generally speaking, IIR filters are minimum phase. However more importantly they are much more efficient at filtering, particularly at low frequencies where an equivalent FIR filter would require 1000’s of taps. I recommend you read:
>> If so, how am I supposed to differentiate them from mixed responses e.g. room responses?
It depends on why you want to differentiate between minimum phase and mixed or non-minimum phase components.
The FIR Creator workflow lets you do whatever you like. You can push the measurement towards target with only minimum phase filters (using the IIR, FIR Mag Adjust and FIR Auto Mag) or you can use additionally use the FIR Phase Adjust and FIR Auto Phase tabs to correct for non-minimum phase components. It’s completely up to you. We don’t advocate any particular method. We just provide the tools to enable you to try whatever approach you like.
EA Support4 September 2023 at 6:43 pm #702
I can’t explain why 1.0.2 stopped working for you.
We’ve checked the builds. SM version 1.0.0, 1.0.2 and 1.0.3 were built with Qt 6.4, which does support Catalina. SM 1.0.4 and 1.0.5 were built with Qt 6.5.x, which doesn’t support Catalina. We always intended SM to be on Qt 6.5.
SM version 1.0.4 (built with Qt 6.5.1) fixed a problem with an intermittent glitch in the transfer function which all users need. Version 1.0.5 addressed problems in Qt itself and was built with Qt 6.5.2.
All this to say, we can’t go back and build 1.0.5 with older versions of Qt, so unfortunately you’ll need to either update macOS, or find a MS Windows PC.
EA Support31 August 2023 at 2:33 pm #696
The data across the formats is effectively identical, with some caveats for WAV.
CSV(row) has FIR coefficients formatted as text in a row.
0.0010, -0.07, 1.0, 0.03 …..
CSV(column) has FIR coefficients formatted as text in a column
WAV uses 24 bit WAV files to store the coefficients. WAV does quantize the data to 24 bit depth, but still has an effective range of > 130 dB. WAV also clips coefficients that are > 1.0 or < -1.0. But “FIR Creator EX” will give a warning if your FIR filter coefficients are outside -1.0 to 1.0 and you are trying to save as WAV.
I’d recommend using one of the CSV formats.
EA Support23 August 2023 at 8:01 am #690
The final length depends on your measurement and exactly what filtering is needed to achieve your goals. Just try it in FIR Creator EX. On the Export tab you can adjust the filter length and delay to where the filter is working the way you want, and you can see the result applied to your loudspeaker measurement. If you’re unsure how to use the workflow, take a look at the videos and tutorials on https://eclipseaudio.com/fir-filter-design-tutorial/.