[musicreo]

These are not done with impulcifer but in the end the principle is always the same.

The CIPIC files are included in HESUVI https://www.ece.ucdavis.edu/cipic/spatial-sound/hrtf-data/
I think this is also included: http://recherche.ircam.fr/equipes/salles/listen/download.html
Here are some dummy head recordings, that are not included: https://audiogroup.web.th-koeln.de/
Another database: https://www.york.ac.uk/sadie-project/database.html

 

[johnn29]

Absent of any reflections the center channel would actually be pretty hard to locate.

Been happily using Impulcifer for months now. Risen to 50% of my viewing/listening. Next door house is a construction site and my cellar below me is having to have industrial extraction fan to cope with the humidity due to some construction there too. A crap show for watching high dynamic range movies - noisy as hell.

Bose 700 ANC on, Impulcifer processing - silent bliss. Can't get over it.

 

[halcyon]

Absent of any reflections the center channel would actually be pretty hard to locate.

Interesting. In real life, with natural sound sources (without visual cues) this is fairly easy.

Then again, IRL again, the right behind sound direction can be somewhat confusing (without head movement). But tons more accurate than any of the Virtual Sound Processing algos are able to achieve.

IMHO, almost all of the Virtual Surround Processing algos fail abysmally with behind the listener sound localization.

Then again, in my own experience even inside a completely dead full size anechoic chamber, this cone cone of confusion is comparatively smaller again IRL (compared to Virtual Surround).

Why are the virtualization algos not able to achieve the same level of sound source localization for behind the listener sources? Is it all down to (lack of) head movement?

 

[musicreo]

How is the mic calibration (left-mic-calibration.csv/right-mic-calibration.csv) processed into the final result? So far I don't hear difference?

 

[jaakkopasanen]

How is the mic calibration (left-mic-calibration.csv/right-mic-calibration.csv) processed into the final result? So far I don't hear difference?

Mic calibration is not supported. There's only calibration for a single room measurement mic. That's baked into the impulse responses by eq when doing room correction.

 

[jaakkopasanen]

Interesting. In real life, with natural sound sources (without visual cues) this is fairly easy.

Then again, IRL again, the right behind sound direction can be somewhat confusing (without head movement). But tons more accurate than any of the Virtual Sound Processing algos are able to achieve.

IMHO, almost all of the Virtual Surround Processing algos fail abysmally with behind the listener sound localization.

Then again, in my own experience even inside a completely dead full size anechoic chamber, this cone cone of confusion is comparatively smaller again IRL (compared to Virtual Surround).

Why are the virtualization algos not able to achieve the same level of sound source localization for behind the listener sources? Is it all down to (lack of) head movement?

The localization perception is very much an individual thing. I fail to get frontal localization with all virtual surround systems except when the BRIR is my own. I think the majority leans towards front localization being harder than rear localization but I have no data to back this up. Virtual surround with headphones have other perceptual challenges as well because when you wear headphones you are aware of it and even the perfect matching BRIR can be confusing when you consciously know that you're listening to headphones. Another problem is the potential mismatch between room acoustics of the BRIR and the room you're listening in.

 

[pfzar]

Localizations: Externalization, front back/ up down
Localization.

 

[outerspace]

IMHO, almost all of the Virtual Surround Processing algos fail abysmally with behind the listener sound localization.

Sit on chair, close your eyes and ask someone to snap fingers in 0.1-1 m distance around you. Try to point out were sound is come from. You will realize you not such precise as you think if eyes are closed. You can confuse sound from front with from back and vice versa. Such experiment you can see here.

 

[johnn29]

pfzar thanks for posting those papers.

The brain uses multiple senses to get an idea where a sound is coming from. In nature i've noticed that head tilting is important. But in small rooms it isn't. If you look at a dog trying to locate a sound - that's an extreme example of what we do.

 

[yosoro]

Traceback (most recent call last):
File "recorder.py", line 260, in <module>
play_and_record(**create_cli())
File "recorder.py", line 206, in play_and_record
min_channels=n_channels
File "recorder.py", line 135, in get_devices
devices = sd.query_devices()
File "C:\Users\17810\Impulcifer\venv\lib\site-packages\sounddevice.py", line 565, in query_devices
for i in range(_check(_lib.Pa_GetDeviceCount())))
File "C:\Users\17810\Impulcifer\venv\lib\site-packages\sounddevice.py", line 565, in <genexpr>
for i in range(_check(_lib.Pa_GetDeviceCount())))
File "C:\Users\17810\Impulcifer\venv\lib\site-packages\sounddevice.py", line 578, in query_devices
name = name_bytes.decode('utf-8')
UnicodeDecodeError: 'utf-8' codec can't decode byte 0xc2 in position 6: invalid continuation byte


Occurs when running recorder

 

[jaakkopasanen]

Traceback (most recent call last):
File "recorder.py", line 260, in <module>
play_and_record(**create_cli())
File "recorder.py", line 206, in play_and_record
min_channels=n_channels
File "recorder.py", line 135, in get_devices
devices = sd.query_devices()
File "C:\Users\17810\Impulcifer\venv\lib\site-packages\sounddevice.py", line 565, in query_devices
for i in range(_check(_lib.Pa_GetDeviceCount())))
File "C:\Users\17810\Impulcifer\venv\lib\site-packages\sounddevice.py", line 565, in <genexpr>
for i in range(_check(_lib.Pa_GetDeviceCount())))
File "C:\Users\17810\Impulcifer\venv\lib\site-packages\sounddevice.py", line 578, in query_devices
name = name_bytes.decode('utf-8')
UnicodeDecodeError: 'utf-8' codec can't decode byte 0xc2 in position 6: invalid continuation byte


Occurs when running recorder

Could you share the full command you used?

 

[yosoro]

python records.py --play =“ data / sweep-seg-FL，FR-stereo-6.15s-48000Hz-32bit-2.93Hz-24000Hz.wav” --record =“ data / my_hrir / headphones.wav”

 

[jaakkopasanen]

Odd. Could you please create an issue here so that we don't spam this thread with the trouble shooting: https://github.com/jaakkopasanen/Impulcifer/issues

 

[reter]

Guys pls i need help!

I've an Asus Xonar U3 and i like its Dolby DH-2, it's spacial but without too much noticeable/annoying reverb, so i started trying to record it's impulse responses to port it to Hesuvi and using another sound card/amp with it but now i'm stuck with something weird:

I read the tutorial in hesuvi page to make it but in the chapter "III. Recording the Impulse Response" point 6 it says to "make sure that you have a single stereo track in the end" but also in the upper chap "II" says to "Select highest number of channels available in the recording channel configuration dropdown and delete to left only two tracks (stereo)", so i did it but when i try to record i get a single impulse response to every channel (in the image)

My question is, how i should record all the impulse responses in only two tracks to make it stereo and import in Hesuvi? i tried to use 2 channels in the channel configuration of Audacity but it gives me an error "error code -9997 invalid sample rate" so i can't record it

What i'm doing wrong? please someone, i really want to import this surround profile that's really different than the DH2 in the Hesuvi HRIR list

 

[Joe Bloggs]

Guys pls i need help!

I've an Asus Xonar U3 and i like its Dolby DH-2, it's spacial but without too much noticeable/annoying reverb, so i started trying to record it's impulse responses to port it to Hesuvi and using another sound card/amp with it but now i'm stuck with something weird:

I read the tutorial in hesuvi page to make it but in the chapter "III. Recording the Impulse Response" point 6 it says to "make sure that you have a single stereo track in the end" but also in the upper chap "II" says to "Select highest number of channels available in the recording channel configuration dropdown and delete to left only two tracks (stereo)", so i did it but when i try to record i get a single impulse response to every channel (in the image)

My question is, how i should record all the impulse responses in only two tracks to make it stereo and import in Hesuvi? i tried to use 2 channels in the channel configuration of Audacity but it gives me an error "error code -9997 invalid sample rate" so i can't record it

What i'm doing wrong? please someone, i really want to import this surround profile that's really different than the DH2 in the Hesuvi HRIR list

It doesn't look like you are doing a loopback recording properly. You should have a cable going from the headphone out on your soundcard to the line in on your soundcard and recording from the line in as you feed the impulses through the different channels. You should not have any option other than stereo for your line in, really.