Why removed reverberation?

[girtabb]

I'm curious if there's a possibility that larger rooms could also come alongside a version with their original reverberation, like the Small broadcast studio

[ShanonPearce]

That shouldn't be a problem. I will include a set of alternative BRIRs without reverberation reductions for the larger rooms.

[girtabb]

Thank you! =)

[NekoAlosama]

ShanonPearce didn't answer the question though. I'm assuming it's because the reverberation makes the sound bad or unrealistic (e.g. you're listening in a small room, but your BRIR comes from a very large room), but we wouldn't know until ShanonPearce includes it.

[ShanonPearce]

The aim of the reverberation reductions was to reduce the amount of distracting late reflections that were present in the larger rooms. As a result there are differences in perceived room size.

I have now added alternative BRIRs without the reverberation reductions for most of the larger rooms so the different versions of the rooms can be compared.

[girtabb]

Hey ShanonPearce! Thanks for uploading these versions. However, I'm a little confused. The versions you added do add more reverberation, but, from what I know, there are supposed to be a lot more reflections I believe (refer to the video, where I compare and use a file from the WDR website's original samples, in which sound only comes from one angle I think). Now I'm just curious if there are more things you consciously do to these samples to create these IR files or if this is unexpectedly happening. Correct me if I understand something incorrectly, I'm not much of an expert yet. Also, I screenshotted the wrong folder (LBS instead of SBS), it's just to maybe clear some confusion as to what kind of files I'm talking about, you might be familiar with those

https://github.com/ShanonPearce/ASH-Listening-Set/assets/131411046/32852184-bc51-4495-a6ed-977beda9c985

[ShanonPearce]

I had another look into my script and there is actually still some windowing active that occurs towards the end of the IR, reducing the overall reverberation time. I will leave the new BRIRs as they are though as removing this windowing will cause a larger mismatch in decay time between the higher frequencies and low frequencies (which are integrated from a separate source and have shorter decay time).

To avoid any further confusion I have updated the naming of the 2 sets of BRIRs to have "Reduced_Reverb" and "More_Reverb"

[girtabb]

So wait, which one has a shorter decay time? The high or the low end? Aren't differences in decay time what's needed to achieve an even wider feeling sound?

[ShanonPearce]

In most of the original room responses, the higher frequencies have shorter decay times than the low frequencies. The challenge I was facing was achieving a natural decay when integrating generalised low frequency BRIRs into the original BRIRs for low frequency extension. Some of the original BRIRs had decay times much longer than the generalised low frequency BRIRs resulting in unusually long decay times in the higher frequencies relative to low frequencies. There is still some reduction in late reflections in the new BRIRs to avoid this problem.

[girtabb]

So if I understand this correctly, you remove some late reflections or apply "windowing" in order to maintain low-frequency extension and relatively similar decay times across the whole frequency response?

[ShanonPearce]

Yes the main intent of the late reflection reduction is to ensure the decay times are relatively consistent across the spectrum and not increasing with frequency after low frequency extension. It isn't as much of an issue for smaller rooms as the low frequency extension tends to integrate nicely with the less reverberant rooms.

[girtabb]

Yeah, I noticed you didn't add the Large broadcast studio. I assume that'd be too much of a hassle to try to make it consistent across the whole spectrum, right?

[ShanonPearce]

Yeah, I noticed you didn't add the Large broadcast studio. I assume that'd be too much of a hassle to try to make it consistent across the whole spectrum, right?

Yeah currently it would be difficult to include the Large broadcast studio in this dataset due to its long reverberation times and a lack of data for low frequency extension. I will look into including it in a future release along with versions of the other large rooms having their original reverberation times. I will first need to improve low frequency extension for rooms containing long reverberation times.

[girtabb]

Thanks a lot! In the end, there shouldn't be a problem if you just add "experimental" HRIRs that lack things like bass extension, right?