[JohnM]

The beta release of V5.01 is now available for download in the Downloads Area. Use the "Drivers" selection box at the top left of the Soundcard preferences to select ASIO then choose the ASIO device, input and output. For most reliable measurements use normal or high latency settings on the ASIO driver.

Changes in this release:

• Added support for ASIO drivers on Windows
• Updated help to add All SPL graph to GraphPanel help
• Tried a tweak to have the big-endian output flag default to on for OS X PowerPC platforms
• Stimulus trace on scope plot is shown at level sweep was made at rather than 100 % FS
• Added sample width selector for dither control on sig gen sine wave generator
• Changed VU meter labelling to show "In" and "Ref In" for input meters with actual input name/channel in tooltip
• Show an error message if user attempts to load an mdat file as a mic/meter or soundcard cal file
• Allow ";" as line comment indicator for imported ASCII files (used by SMAART)
• Added a check on soundcard calibration measurements to warn if measurement appears invalid
• Added impulse response export options to export windowed response or minimum phase response
• Added a warning message after SPL meter calibration if the input level was below -50 dB FS
• Added a MiniDSP-96k equaliser option for 96kHz MiniDSP plug-ins (e.g. for 2x8)
• Added parallel equivalent impedance display for impedance traces
• Bug fix: TS parameters calculation could fail for some measurements
• Bug fix: Problem opening EQ panel with a measurement generated from Trace Arithmetic
• Bug fix: Target trace in EQ window did not include house curve when there were no measurements loaded
• Bug fix: Do not use physical fonts unless default Locale language is en to avoid undisplayable characters
• Bug fix: Export RT60 data as text was broken
• Bug fix: Fill filter traces filled the house curve section of the target if a house curve was loaded
• Bug fix: Force use of decimal point as decimal delimiter in SPL meter SPL value
• Bug fix: Adjusted DCX2496 centre frequencies
• Bug fix: Error when trying to upload TMREQ filters from surround or surround back channels
• Bug fix: Exporting filter impulse response as WAV always exported the set from the currently selected measurement regardless of the selection made in the export dialog
• Bug fix: Path to last imported impulse response was not saved correctly on exit
• Bug fix: Hilbert transform had an error that affected envelope calculation

Please report any issues in this thread.

 

[JohnM]

Re: V5.10 Beta release - ASIO support

Good to hear it seems to be working

I don't have any programmatic control of the latency setting, have to adjust it via the ASIO control panel. REW can query the latency, but whether the detected level is too low is platform dependent so difficult to make a general recommendation. REW does detect and warn about potentially corrupted impulse responses, but small drop-outs may not trigger that.

can you make popup windows (like EQ progress popups) children of their parent windows?

I've fixed that for the EQ progress popup, didn't find any others but let me know if you come across one.

 

[JohnM]

Re: V5.10 Beta release - ASIO support

I get the following error when loading REW after using it successfully in ASIO mode the previous session.

There will be some additional diagnostic info in the log file that has that error, could you attach it or email me a copy please? The files are in your home directory in a folder called REW.

 

[JohnM]

Re: V5.10 Beta release - ASIO support

I get the following error when loading REW after using it successfully in ASIO mode the previous session.

I have uploaded beta 2 which may resolve that error, please try it out and let me know.

 

[JohnM]

Re: V5.10 Beta release - ASIO support

I have uploaded beta 3, which has the following changes:

- the ASIO channels are prefixed by their channel index, that should fix the "3L/3R" problem
- another attempt to stop the LineUnavailableException / input not available on startup after using ASIO
- Add a warning if the ASIO buffer size is not a power of 2 and disable multiple sweeps, that is probably the cause of the odd result with multiple sweeps above
- Allow for driver reset, resync or buffer size change while REW is running
- Suppress errors from the ASIO driver unload on shutdown

 

[JohnM]

Re: V5.10 Beta release - ASIO support

Thanks for that. For the benefit of others following this, 48k would generally be better for measurements unless you need the extra bandwidth to track down ultrasonic tweeter resonances, for example.

Larger buffer size might help with the dropouts, but I have run buffers from 64 samples to 2k samples and don't see much correlation between occasional glitches and buffer length, so it may be more down to whatever else the PC is doing at the time.

 

[JohnM]

Re: V5.10 Beta release - ASIO support

1. My E-MU 0404 ASIO driver can only change time of buffer (ms, not samples). From 2ms to 500ms. On any choise REW inform me abou a "power of 2". What I need to do?

The number of samples is the time multiplied by the sample rate, so (for example) 100ms at 48,000Hz is 0.1 * 48000 = 4,800 samples. Power of 2 sample lengths are 256, 512, 1024, 2048 etc. At 48kHz sampling 1024 samples would be 1024/48000 =
0.0213s = 21.3ms, so a figure around 21ms might give a power of 2 length.

The only feature not available when the length is not a power of 2 is performing multiple sweeps, which is rarely needed anyway, so don't worry if you don't happen to find a time that results in a power of 2.

2. How can I select both channels (left+right) to out via SPDIF of 0404? Now I hear sweep only from left speaker. Native soundcard ASIO driver can't rule the channels.

Select the other channel as the Ref output.

 

[JohnM]

Re: V5.10 Beta release - ASIO support

Why is pow2 necessary for multiple sweeps? Can't you just pad or truncate the recorded data?

It was a hangover from the JavaSound implementation and the synchronous pre-averaging used for multiple sweeps, I've removed the restriction for beta 4 which I have just uploaded.

 

[JohnM]

Re: V5.10 Beta release - ASIO support

Observation #1 : Even though this card is running on USB power ( and therefore limited in it's total count of available input & output channels ) , the drop-down selection menus ( within the preferences window ) don't properly reflect the "hobbled" status of the sound card.

The ASIO driver for the card provides the list of channels available and REW displays them, the drivers tend not to reflect whether the inputs are really available - for example, they will report their channels even if the soundcard is not connected, but there will be an error message if an unavailable channel or soundcard is selected.

Event #1 ; I was surprised to see this notification when I was simply trying to test a loop-back , using the "Calibrate SC" section as a 1K source.

If the "SPL Meter audio input not available" message appears then setting drivers to Java then back to ASIO and reselecting the ASIO inputs/outputs should clear it, though I'm not sure how it might be triggered.

Event #2 ; For a while, I was able to consistently get this notice .

Input data overruns are flagged if old data has not been consumed before new data arrives, I'll try increasing the internal buffers in the next build (beta 5) to see if that helps.

Event #3 ; Stepping back to selecting JAVA/WDM drivers didn't obviously help things ;

The WDM and ASIO drivers may not cope well with switching back and forth.

Event #4 ; Trying the ASIO4ALL drivers only resulted in a worsening of the situation .
ie; One doesn't want to see 2 simultaneous "Notices" like this / this led to a forced quit situation ( then a reboot ).

ASIO4ALL has been well behaved for me. If input overrun messages appear they have to be acknowledged to continue.

Event #5 ; Under some specific circumstances ( opening , closing windows ) I can get the ASIO outputs to lockup and go into a rhythmic chatter .
- I've uploaded a "zipped" WAVE file of what this sounds like .

Don't think it will top the iTunes downloads chart I've not had any problems like that so if you find a way to reliably reproduce it let me know.

 

[JohnM]

Re: V5.10 Beta release - ASIO support

I have uploaded Beta 5, which only flags input data overruns if they could affect a measurement result. Some drivers (e.g. Tascam US-144) block the REW audio init thread when they first start up (for a few hundred ms) but start sending data during that time, which caused spurious overruns on startup.

Drivers also seem to vary regarding whether they inform clients of buffer size/latency changes - they should do, ASIO4ALL does, but the Tascam driver doesn't and if REW has already been run with the old settings it will fail after silent changes to buffer sizes. It is safest to toggle the REW driver setting to Java then back to ASIO after making a change in the ASIO control panel.

 

[JohnM]

Re: V5.10 Beta release - ASIO support

It's clear, but in analog mode I can use Y-splitter and feed both channels of receiver. For fiberchannel interface it's not applicable =)
Any solution?

Do as I said in reply to your previous post, select the other channel you want the signal to appear on as the Timing Reference Output in the REW soundcard preferences, then the test signal will appear on both.

 

[JohnM]

Re: V5.10 Beta release - ASIO support

I'm getting a lot of visual corruption on the EQ window.

Thanks, it has been a while since I ran the utility to spot component updates off the Event Despatch Thread, that picked up a status bar update that wasn't being done from the EventQueue - I've fixed that for the next beta, it might solve that problem.

 

[JohnM]

Re: V5.10 Beta release - ASIO support

I have uploaded beta 6 with the following changes:

• Made ASIO input buffering more efficient
• Bug Fix: Ohms scale appeared on RHS of SPL overlay even with no impedance measurements loaded
• Bug fix: Possible fix for corruption of EQ window when automatically setting filters

 

[JohnM]

Re: V5.10 Beta release - ASIO support

Please can you make the ouput and input selections more independant.

Unfortunately that would be quite difficult to implement. Do the Java drivers not work for the output?

 

[JohnM]

Re: V5.10 Beta release - ASIO support

I can put this new ASIO Beta(6) into an @ 80% failure rate / getting the Impulse Peak to show at the wrong place .

Is that using ASIO or Java drivers Earl? Does sample rate make a difference?

 

[JohnM]

Re: V5.10 Beta release - ASIO support

Beta 6 known issue: When using multiple sweeps with Java drivers only a single sweep is generated. This will be fixed in the next release.

 

[JohnM]

Re: V5.10 Beta release - ASIO support

- The faster the sweep ( along with the higher the sweeps start frequency ), the more the certainty, that the IR will be wrong .

Beta 7 has some improvements that may fix that

 

[JohnM]

Re: V5.10 Beta release - ASIO support

I have a work-around of not using my USB DAC, but that is not ideal.

Have you tried installing ASIO4ALL? That should provide ASIO interfaces for all devices.

 

[JohnM]

Re: V5.10 Beta release - ASIO support

I'm still getting the EQ window corruption with beta7.

It's not reliably reproducable, but the trigger seems to be the redraw event that happens after I minimize the main REW window, and then show it again. I just made it happen by clicking on the REW taskbar entry to minimize it, and again to show it again (the EQ window was maximized the whole time).

That sounds more like an issue between the JRE and your graphics card. JRE 6 introduced a few of those as the underlying graphics interfaces were rewritten. Updating the graphics card drivers often fixes it, but if you are already up to date you may be out of luck, but could be worth reporting it to Oracle.

 

[JohnM]

Re: V5.10 Beta release - ASIO support

a loopback during measurement is used to determine the latency through the system right? With ASIO that's presumably no longer required, as you can query the latency accurately from the driver - or is that not enough?

REW's loopback reference connection allows it to eliminate all delays in the reference chain, which may include more than the driver, such as internal buffering within REW and any processing delays in the chain.