GSonic Reference – Free Stereo Room Correction Tool (Measurement + FIR Export)

[kharan]

GSonic Reference v1.0.25 released.

• Report.txt is now written next to the filters: transition frequency, speaker roll-off, stereo balance, headroom, and a band-by-band check of how left and right sum at the listening position. If your room cancels mono bass at some frequency, it shows up here.
• Bass Phase Alignment, optional and off by default (Generate > Advanced). Rotates left and right in opposite directions below 100 Hz so mono bass sums in phase at the listening position. Pure allpass, no level change, added group delay capped by a slider. In my room it filled a narrow 80 Hz null with no image shift; the audible gain was small, so try it only if the report flags a cancellation.
• Phase correction now uses a frequency-dependent window (6 cycles, 8 ms maximum) instead of a fixed 4.5 ms, so the tweeter is windowed anechoically and low crossovers get more resolution. Correction is faded where the three measurement positions disagree or inside deep notches, and whole turn phase laps through nulls are no longer treated as speaker phase.
• Bass correction is designed with a 2 Hz minimum bandwidth so the 32768-tap filter reproduces it exactly. Previously the narrowest cuts could be off by up to 1.5 dB after export.
• Exported filters never exceed 0 dB at any frequency. Heavily limited tracks may still need a couple of dB of headroom in the convolver, as before, by design.
• Sub-sample alignment of the spatial average had its sign reversed; fixed. Small effect on treble phase only.
• The measurement prompt now suggests placing the two side positions about an ear's width from MLP.

Enjoy!

So new measures are needed, or you can use an existing set of measures and generate new corrections?

 

[OCA]

So new measures are needed, or you can use an existing set of measures and generate new corrections?

You can use your old measurements. All changes are related to filter generation but a new measurement takes a few minutes anyway.

 

[GGroch]

OCA, just wanted to say thanks for this. I hated REW enough that I stopped EQing my desktop system. GSonic is perfect for me. Plus, I did not know about Convolution filters and their superiority to text/parametric filters. It took a few minutes for me to learn how to integrate them into PEACE/AutoEQ, but it works great. Thanks again for your work.

 

[YouDontHaveEnoughBass]

amazing tool thank you very much

quick question, is it ok to take the 3 measurements across a listening sofa (so 1st left side of sofa, second in the middle, third on the right)?

 

[StabMe]

Yeah, by far the best correction i was able to get ever.

Glad you got back to work on Gsonic Reference. Maybe it will make release of the full version with full MIMO closer

 

[OCA]

amazing tool thank you very much

quick question, is it ok to take the 3 measurements across a listening sofa (so 1st left side of sofa, second in the middle, third on the right)?

Yes, min phase correction will end earlier since the common filter area will shrink but the process is otherwise the same.

 

[DenonMate]

Yeah, by far the best correction i was able to get ever.

Glad you got back to work on Gsonic Reference. Maybe it will make release of the full version with full MIMO closer

Just preparing steps to get to the olymp. GSR, GSI, GSEvo32, GS…. (MIMO)

 

[NotInYourHead]

GSonic Reference v1.0.25 released.

• Report.txt is now written next to the filters: transition frequency, speaker roll-off, stereo balance, headroom, and a band-by-band check of how left and right sum at the listening position. If your room cancels mono bass at some frequency, it shows up here.
• Bass Phase Alignment, optional and off by default (Generate > Advanced). Rotates left and right in opposite directions below 100 Hz so mono bass sums in phase at the listening position. Pure allpass, no level change, added group delay capped by a slider. In my room it filled a narrow 80 Hz null with no image shift; the audible gain was small, so try it only if the report flags a cancellation.
• Phase correction now uses a frequency-dependent window (6 cycles, 8 ms maximum) instead of a fixed 4.5 ms, so the tweeter is windowed anechoically and low crossovers get more resolution. Correction is faded where the three measurement positions disagree or inside deep notches, and whole turn phase laps through nulls are no longer treated as speaker phase.
• Bass correction is designed with a 2 Hz minimum bandwidth so the 32768-tap filter reproduces it exactly. Previously the narrowest cuts could be off by up to 1.5 dB after export.
• Exported filters never exceed 0 dB at any frequency. Heavily limited tracks may still need a couple of dB of headroom in the convolver, as before, by design.
• Sub-sample alignment of the spatial average had its sign reversed; fixed. Small effect on treble phase only.
• The measurement prompt now suggests placing the two side positions about an ear's width from MLP.

Enjoy!

All great QoL updates, and coincidentally when I have just rearranged my setup. Thanks OCA

 

[piaseczek]

Bass Phase Alignment, optional and off by default (Generate > Advanced). Rotates left and right in opposite directions below 100 Hz so mono bass sums in phase at the listening position. Pure allpass, no level change, added group delay capped by a slider. In my room it filled a narrow 80 Hz null with no image shift; the audible gain was small, so try it only if the report flags a cancellation.

It seems that even with the slider at 5ms, new option solved my SBIR problems at 73hz. Thank you!

 

[NotInYourHead]

It seems that even with the slider at 5ms, new option solved my SBIR problems at 73hz. Thank you!

Did increasing the slider have any other significant effect for you? Would it change how bass is felt subjectively?

 

[YouDontHaveEnoughBass]

i gotta say this things tremendous; i keep flicking it on and off (using the bypass mode) while listening to music or gaming and the spacial improvements incredible - done it both with my living room setup and my girlfriends desktop setup and its insane for both

my girlfriends not an audiophile at all and doesnt particularly care about the tonal differences, says she can tell the difference but not necessarily say which ones better, but really likes the spacial qualities when she's playing games she loves how everything 'locks in'

thanks again for the great tool

 

[landco1]

GSonic Reference v1.0.25 released.

No 2.1 support (subwoofer)

 

[NotInYourHead]

No 2.1 support (subwoofer)

This is for 2.0 setups only.

 

[YouDontHaveEnoughBass]

would it be possible to override the auto calculated roll-off? im trying to use this with another setup but, due to a deep cancellation in the room, it keeps thinking the speakers roll off at 120hz 18db which they certainly dont

would be very useful to be able to manually specify this

 

[NotInYourHead]

would it be possible to override the auto calculated roll-off? im trying to use this with another setup but, due to a deep cancellation in the room, it keeps thinking the speakers roll off at 120hz 18db which they certainly dont

would be very useful to be able to manually specify this

Funny, I was about to ask @OCA the same question. For me it was a null at 70Hz that tripped GSonic and set both L/R rolloff to 70Hz even though the bass extension is good to 40Hz. I can attach the measurement wavs if he wants to repeat the problem.
I tried with 1.0.24 and 1.0.25, both reaches the same rolloff.

 

[piaseczek]

Did increasing the slider have any other significant effect for you? Would it change how bass is felt subjectively?

I only checked the reports and compared the setting 5ms to 10ms. The lower value the better. Listening is ahead of me.

 

[john61ct]

The width of an ear as in say under 10mm??

Or something like a skull's width?

 

[OCA]

GSonic Reference v1.0.26 released.

Fixes from your feedback

• Bass roll-off detection no longer mistakes a room cancellation for the speaker's roll-off. A candidate roll-off below which the response climbs back up is refused, and the log says which null was refused. Previously a deep null at 70-120 Hz could switch bass correction off below it.
  
  
• Bass Roll-off override (Generate > Advanced): leave it on Auto, or set the corner by hand from 20 to 120 Hz when you know your speakers better than the measurement does.
  
  
• Bass polarity is now decided from the full measured response instead of a short window, so it no longer varies between versions in rooms where the low-frequency phase sits near 90 degrees. Report.txt shows the decision. If your bass phase looked different between v1.0.24 and v1.0.25, this settles it.

Rooms not affected by these cases get filters identical to v1.0.25.

 

[Bhavik Patel]

GSonic Reference v1.0.26 released.

Fixes from your feedback

• Bass roll-off detection no longer mistakes a room cancellation for the speaker's roll-off. A candidate roll-off below which the response climbs back up is refused, and the log says which null was refused. Previously a deep null at 70-120 Hz could switch bass correction off below it.
  
  
• Bass Roll-off override (Generate > Advanced): leave it on Auto, or set the corner by hand from 20 to 120 Hz when you know your speakers better than the measurement does.
  
  
• Bass polarity is now decided from the full measured response instead of a short window, so it no longer varies between versions in rooms where the low-frequency phase sits near 90 degrees. Report.txt shows the decision. If your bass phase looked different between v1.0.24 and v1.0.25, this settles it.

Rooms not affected by these cases get filters identical to v1.0.25.

Unable to open the link. Anyone else is facing same issue?

 

[OCA]

Unable to open the link. Anyone else is facing same issue?

Fixed now, thanks.