Re: Any errors?
The software installs, then at the Splash screen, a Windows applet comes on and says that "The software has stopped working".
I uninstalled Ver 1.2 and then tried to reinstall ver 1.1.6, the same installation that worked before. Now I can't seem to install any version, they all stop at the splash screen - "configuring components" I believe.
I will use any version that works on 64-bit. Most new computers sold here in the US are using the 64-bit version because it is so much faster and more stable.
I am not at the moment connected to the network - is this required to install?
askbojesen said:
The software installs, then at the Splash screen, a Windows applet comes on and says that "The software has stopped working".
I uninstalled Ver 1.2 and then tried to reinstall ver 1.1.6, the same installation that worked before. Now I can't seem to install any version, they all stop at the splash screen - "configuring components" I believe.
I will use any version that works on 64-bit. Most new computers sold here in the US are using the 64-bit version because it is so much faster and more stable.
I am not at the moment connected to the network - is this required to install?
Re: Re: Block Export to textfile
(base 2 log(order A/order B))/(sweep rate in octaves per second)
I think Ask's software shows the locations of the distortion order peaks though?
On the general issue of using sweep measurements for distortion, the main virtue of the log sweep is the ease with which distortion products can be excluded rather than analysed. Whilst the IR can be analysed to provide a very quick measure of distortion, there is not enough energy in the stimulus to accurately measure low levels of distortion even with very long sweeps - and if you are not measuring in an anechoic environment you need to use a long sweep so that the sweep rate is low enough to keep sufficient separation between the harmonis. A stepped sine approach will be much more accurate, but of course takes a lot longer.
The time separation between harmonic orders is given bygedlee said:
(base 2 log(order A/order B))/(sweep rate in octaves per second)
I think Ask's software shows the locations of the distortion order peaks though?
On the general issue of using sweep measurements for distortion, the main virtue of the log sweep is the ease with which distortion products can be excluded rather than analysed. Whilst the IR can be analysed to provide a very quick measure of distortion, there is not enough energy in the stimulus to accurately measure low levels of distortion even with very long sweeps - and if you are not measuring in an anechoic environment you need to use a long sweep so that the sweep rate is low enough to keep sufficient separation between the harmonis. A stepped sine approach will be much more accurate, but of course takes a lot longer.
Re: Re: Re: Block Export to textfile
I don't think that seperation is an issue, but yes, the energy level might be. The point is that I have not, as yet, seen any technique that yields reliable acoustic measurements of harmonic orders above the third. The key word here being "reliable" - lots of them print numbers. The second and third are basically inaudible so they are kind of uninteresting and the interesting ones seem to be extremely hard to get. I was hoping that this software could help with that problem, but I guess not.
I would tend to agree that steped sine has some benefits and I may look at doing this myself. With steped sine you could lock into the frame rate for each sine wave and analyze the harmonics synched to the window for very high resolution. That still doesn't mean that reliable data for the higher harmonics will be obtained. It could be that these harmaonics are not even stable in the system itself.
JohnPM said:
I don't think that seperation is an issue, but yes, the energy level might be. The point is that I have not, as yet, seen any technique that yields reliable acoustic measurements of harmonic orders above the third. The key word here being "reliable" - lots of them print numbers. The second and third are basically inaudible so they are kind of uninteresting and the interesting ones seem to be extremely hard to get. I was hoping that this software could help with that problem, but I guess not.
I would tend to agree that steped sine has some benefits and I may look at doing this myself. With steped sine you could lock into the frame rate for each sine wave and analyze the harmonics synched to the window for very high resolution. That still doesn't mean that reliable data for the higher harmonics will be obtained. It could be that these harmaonics are not even stable in the system itself.
Re: Re: Re: Re: Block Export to textfile
The separation becomes a problem in real (rather than anechoic) environments because the response of the harmonic order can last longer than the separation between harmonics, so the orders start to interfere with each other.
Very reliable results can be obtained by using a sine generator that adjusts the stimulus frequency so that an exact multiple of periods fits within the FFT block length used for the analysis, that ensures the harmonics are centred on FFT bins. If only a few harmonics are of interest then you can simply use a Goertzel filter for each harmonic.
I believe Ivo Mateljan's STEPS software provides very accurate distortion measurements using stepped sine.
gedlee said:
The separation becomes a problem in real (rather than anechoic) environments because the response of the harmonic order can last longer than the separation between harmonics, so the orders start to interfere with each other.
Very reliable results can be obtained by using a sine generator that adjusts the stimulus frequency so that an exact multiple of periods fits within the FFT block length used for the analysis, that ensures the harmonics are centred on FFT bins. If only a few harmonics are of interest then you can simply use a Goertzel filter for each harmonic.
I believe Ivo Mateljan's STEPS software provides very accurate distortion measurements using stepped sine.
64 bit version
64 bit version
http://www.holmacoustics.com/holmimpulse.php
http://www.holmacoustics.com/downloads/HOLMImpulse/HOLMImpulse-1.2.0.0-x64.msi
64 bit users: Please report back is this version works
gedlee said:
64 bit version
http://www.holmacoustics.com/holmimpulse.php
http://www.holmacoustics.com/downloads/HOLMImpulse/HOLMImpulse-1.2.0.0-x64.msi
64 bit users: Please report back is this version works
N'th Pre-Impulses timewindows
Yes I show them with indicators - I use them as sanity check
Ok. Lets the n-th time window:
Let's say P[n] defines the sample number for n'order hd-pre-impulse
then the window I use is
From: 2/3*P[n] + 1/3*[n-1]
To: 2/3*P[n] + 1/3*[n+1]
I use these values based on comparison with steady state measurements.
It is a compromise of course - but with this window I get the correct normalization compared with the steady state measurements.
I invite you to do the same comparison on your speakers
(Yes of course we must do this is near-field to avoid the echoes of the room to go outside the above window) Like you correctly say is a problem.
JohnPM said:
Yes I show them with indicators - I use them as sanity check
Ok. Lets the n-th time window:
Let's say P[n] defines the sample number for n'order hd-pre-impulse
then the window I use is
From: 2/3*P[n] + 1/3*[n-1]
To: 2/3*P[n] + 1/3*[n+1]
I use these values based on comparison with steady state measurements.
It is a compromise of course - but with this window I get the correct normalization compared with the steady state measurements.
I invite you to do the same comparison on your speakers
(Yes of course we must do this is near-field to avoid the echoes of the room to go outside the above window) Like you correctly say is a problem.
Re: Re: Re: Re: Re: Block Export to textfile
I understand about synchronizing the time based window to the sine wave. I've done this before myself. Its easy to do for electronics, not as easy to do with acoustics and the need to window the data.
Do you have a link for that software?
JohnPM said:
I understand about synchronizing the time based window to the sine wave. I've done this before myself. Its easy to do for electronics, not as easy to do with acoustics and the need to window the data.
Do you have a link for that software?
STEPS is part of the ARTA package, although ARTA, STEPS and LIMP are separate programs. All available here: http://www.fesb.hr/~mateljan/arta/download.htm
Hi Ask
Thanks for the 64-bit code, it installs now, but we are back to the same problem from version 1.4. There is no input listing under devices and whatever is selected as the output device crashes the program. This is exactly what it did before, but you seemed to have fixed it in version 1.6 because it ran for a time. Its back now.
Thanks for the 64-bit code, it installs now, but we are back to the same problem from version 1.4. There is no input listing under devices and whatever is selected as the output device crashes the program. This is exactly what it did before, but you seemed to have fixed it in version 1.6 because it ran for a time. Its back now.
Re: Re: Re: N'th Pre-Impulses timewindows
Why would you care? Your ears don't know the difference.
I would also be very concerned with a nearfield distortion measurement on a waveguide. I'm not sure what it would mean. The far-field is where we listen and its the distortion there that matters (if it matters at all, of course).
soongsc said:
When doing this in far field, how would one distinguish the source causing harmonics?
Why would you care? Your ears don't know the difference.
I would also be very concerned with a nearfield distortion measurement on a waveguide. I'm not sure what it would mean. The far-field is where we listen and its the distortion there that matters (if it matters at all, of course).
Versions: 32 bit, SSE, 64 bit
Thank you very much for testing this.
The amount of RAM being used depends on how many measurements you have in your 0-49 slots and your signal length.
I'll just repeat myself:
SSE2 - The SSE2 version is running slightly faster.
Does any of you NOT use the SSE2 version?
Reformulated questions:
Have any of you got the SSE2 version to crash while the other did not crash?
Key said:
Thank you very much for testing this.
The amount of RAM being used depends on how many measurements you have in your 0-49 slots and your signal length.
I'll just repeat myself:
SSE2 - The SSE2 version is running slightly faster.
Does any of you NOT use the SSE2 version?
Reformulated questions:
Have any of you got the SSE2 version to crash while the other did not crash?
Thanks for the new version and the nice changes.
I really like the level control by dBFS. I'm now running at -3 or -6, so no fear of clipping anything. The new time thing is cool, too - for driver offsets and delays. Very handy.
One thing I would rather see is the mic and DAC calibration moved back to the Device & Signal tab. The wave outputs could go on the data analysis tab. At least that works best for me.
As for the SSE2 - how do we know? I don't remember what I installed, and the help panel does not seem to say.
Keep up the great work!
I really like the level control by dBFS. I'm now running at -3 or -6, so no fear of clipping anything. The new time thing is cool, too - for driver offsets and delays. Very handy.
One thing I would rather see is the mic and DAC calibration moved back to the Device & Signal tab. The wave outputs could go on the data analysis tab. At least that works best for me.
As for the SSE2 - how do we know? I don't remember what I installed, and the help panel does not seem to say.
Keep up the great work!