REW macOS DMG (includes private Java 8 JRE) V5.20 RC11

Changes in V5.20 RC11:

• Limit the impedance graph Q overlay value to 999 and DF value to 200 %
• Bug fix: Impedance cal file values were written with insufficient precision (unintended side effect of change to allow unit selection for exports)
• Bug fix: Impedance reference cal measurement was limited to 2 Hz even if the axis limit was lower
• Bug fix: Trying to open mdat files directly from web pages with Firefox could fail with a file not found message

Changes in V5.20 RC10:

This build adds an option on the Measure dialog to make a series of SPL measurements by setting the total number required in the Measurements count. A typical use would be when measuring polar responses. The Measure dialog delay is applied before each measurement, to provide time to adjust the speaker turntable, and there is an option to pause the sequence. Measurement names are updated for each measurement according to the Measure dialog settings, for polar responses that would usually mean adding a number after each measurement that increments by the angular interval.

Other changes in this build:

• Allow SPL * impedance or SPL / impedance in trace arithmetic operations, normalising the impedance to the ref resistance for power figures
• Added a units selection on the export measurement as text dialog
• Added an equaliser entry for PerListen Audio
• Added an RTA appearance option to show the peak SPL for tone bursts when in spectrum mode
• Increased tolerance to delays before the sweep in offline sweep recordings
• For impedance measurements added Q (for positive phase) or DF (for negative phase) to the impedance series/parallel overlay box
• Improved the Windows & Linux file chooser column width handling, select details view by default, show current path in dialog title
• Reduced Scope memory use for large time spans
• On Windows and MacOS use the dark background by default if the OS is set to use a dark theme for apps. If the REW theme is changed using the View preference the choice will be respected regardless of OS settings.
• Recommend the 32-bit build for machines running Intel(R) Celeron(R) J1XXX or N2XXX processors due to crashes when running the 64-bit build
• If imported data looks to be accurately linearly spaced in frequency at less than 1 Hz spacing import it as is without resampling

Bug fixes:

• Bug fix: Custom filtered periodic noise could return a zero-valued sequence for some filter slopes
• Bug fix: .req files should have their own "EQ" icon
• Bug fix: The alignment tool would show an error message when aligning phase if the phase traces had not been displayed
• Bug fix: The minimum valid frequency for a multi-input average could be lower than the minimum valid frequency for the measurements that were averaged

Changes in V5.20 RC9:

A collection of small bug fixes with a couple of minor tweaks.

• Use the last req file notes as the default entry when saving a req file
• Allow saved filenames to include commas
• Speed up saving mdat files to network storage
• Changed "multi-channel capture" to "multi-input capture" to avoid confusion
• Bug fix: If the minimum phase impulse is visible on the Impulse graph when adjusting t=0 it will move with the time axis
• Bug fix: The response could be stretched along the frequency axis after applying a frequency dependent window
• Bug fix: Changing the distortion graph X axis between generator and input for a stepped level measurement shifted the plot
• Bug fix: The Linearity trace for a stepped level measurement was not drawn correctly if the Y axis was not SPL on startup
• Bug fix: The minimum phase impulse response data start time could be offset by a few us when using a timing reference, causing the plot to be shifted a few us along the time axis
• Bug fix: Impedance measurements did not go below 2 Hz even if the axis start was set lower
• Bug fix: Periodic Noise sequences exported to WAV could overflow at rms levels above approx -6 dBFS. Note that using such high rms levels will cause clipping and affect the frequency response.
• Bug fix: Clearing and then reloading a USB mic cal file did not load the sensitivity data until the next restart or input change
• Bug fix: Removed spurious dBFS label on SPL meter levels bar
• Bug fix: EQ filters panel frequency figure sometimes showed more decimal places than required
• Bug fix: On Win 2003 an exception could occur if a WAVE volume control was found
• Bug fix: Invert impulse should not affect the minimum phase response
• Bug fix: Estimate IR delay uses the current IR windows (previously used fixed window lengths) so can also be applied to FDW data
• Bug fix: On macOS saving all measurements with one existing mdat loaded would prompt with the whole path of the existing mdat in the filename box instead of just the filename

Minor changes and bug fixes in V5.20 RC8b:

• Change stepped sine start and end frequencies to spinners with 1 Hz steps. Actual test frequencies will be the preferred values at the chosen measurement PPO that are within the span.
• Allow stepped sine start level to be greater than end level
• Soundcard calibration uses longer sweeps if the frequency axis start value is less than 2 Hz
• Changed the shortcut for removing smoothing to Ctrl+0 as the previous Ctrl+shift+0 shortcut is intercepted by the OS on some Windows versions including Win 10
• Continue to show the impulse response window extent indicators when the t=0 adjust dialog is being used
• Show a confirmation message when measurements have been generated from a room sim response
• Added a view preference to control whether changes in the measurements list are animated
• Bug fix: Distortion graph volts X axis for stepped level measurements could zoom in too much and break the graph
• Bug fix: RTA levels panel figures excluded data above 95% of Nyquist if the distortion LP was not being used
• Bug fix: Some plots using dBFS as Y axis did not update immediately when the Full scale sine rms is 0 dBFS preference was changed

Minor changes and bug fixes in V5.20 RC8:

• Change stepped sine start and end frequencies to spinners with 1 Hz steps. Actual test frequencies will be the preferred values at the chosen measurement PPO that are within the span.
• Allow stepped sine start level to be greater than end level
• Soundcard calibration uses longer sweeps if the frequency axis start value is less than 2 Hz
• Changed the shortcut for removing smoothing to Ctrl+0 as the previous Ctrl+shift+0 shortcut is intercepted by the OS on some Windows versions including Win 10
• Continue to show the impulse response window extent indicators when the t=0 adjust dialog is being used
• Show a confirmation message when measurements have been generated from a room sim response
• Added a view preference to control whether changes in the measurements list are animated
• Bug fix: Distortion graph volts X axis for stepped level measurements could zoom in too much and break the graph
• Bug fix: RTA levels panel figures excluded data above 95% of Nyquist if the distortion LP was not being used
• Bug fix: Some plots using dBFS as Y axis did not update immediately when the Full scale sine rms is 0 dBFS preference was changed

V5.20 RC7

Thank you to the folk who have reported bugs or error messages over the last month, it really is very helpful. This build has the following bug fixes:

• Bug fix: Array index out of bounds could occur on the spectrogram
• Bug fix: scanning for 8c could fail if the network interface returned a "no route to host" error
• Bug fix: The RTA N result could be incorrect when using manual fundamental
• Bug fix: Pasted target settings were not saved with the measurement
• Bug fix: Processing WAV files in the RTA could cause an exception when REW is configured for multiple inputs
• Bug fix: Multi-channel capture did not work properly when using a loopback as timing reference

Despite my intention to minimise feature additions ahead of the 5.20 release I have made a number of changes and feature additions in this build, mainly around the RTA.

RTA changes:

• Expand the allowed distortion LP and HP settings so both allow 0 .. 99,999 Hz so arbitrary ranges may be selected
• The RTA search for a fundamental is restricted to the distortion HP to LP span
• REW now uses a notch filter compliant with the AES17-2015 definition for THD+N results, bandwidth is 1 octave. AES-17 specifies its "standard notch filter" rather loosely, requiring a Q range of between 1.2 and 3 which is a bandwidth of between 1.17 and 0.48 octaves. 1 octave gives results close to those obtained with commercial hardware analysers.
• Clarified in the RTA help that the rms input level (and the A and C weighted values) are calculated over the range specified by the distortion LP and HP settings, if either or both are enabled, regardless of the state of the Show distortion button. To make that more apparent the fading out of regions outside the LP and HP span will always be applied, previously it was only applied when the Show distortion button was selected. That allows the distortion HP and LP to be used to obtain the rms level over arbitrary frequency spans.
• The level of the fundamental in the RTA distortion panel is shown in the current axis units as well as in dBFS
• The N (noise and non-harmonic distortion) figure in the RTA distortion panel is shown in the current axis units
• Noise-related parameters in the RTA distortion panel are shown orange when using coherent averaging as they are not valid
• Allow frequencies down to 1.0 Hz for stepped sine measurements
• Added 96 kHz as an end frequency for stepped sine measurements

Other changes

• There is a new Analysis preference to apply cal files to distortion results, both log sweep and stepped sine. The default is not to apply cal files, for consistency with previous builds. Applying the cal files provides more accurate results in regions where the fundamental or harmonics are affected by interface roll-offs but are above the noise floor, but it also boosts the noise floor in those roll-off regions which must be borne in mind when looking at the results.
• There is a new View preference to set the frequency axis start value between 0.1 Hz and 20 Hz (previously the lower limit was 2 Hz). Some applications, such as examining low frequency noise behaviour, need very extended low frequency results (with correspondingly large FFT or sweep lengths and extended range cal files). Conversely, some applications may never need to see frequency values below 20 Hz. The new preference allows the axis limit to be adapted to the needs of the application.
• A set of preset options have been for the full scale sine Vrms entries on the generator, RTA and Scope, accessed by clicking the small triangle in the upper left corner of the entry field
• The numeric value shown on output level meters is in the units that have been selected on the generator (dBu, dBV, Volts or dBFS)
• The numeric value shown on input level meters is in the units selected by right-clicking on the level meters panel (dB SPL, dBFS, dBu, dBV, dBW, Volts or Watts)
• The slice interval is shown in the waterfall info label

Changes in V5.20 RC6

Continuing to focus on remaining bug fixes ahead of the final V5.20 release, but a few small features have been added:

• Added an equaliser entry for the AU N-Band EQ
• Allow TS params driver area entries down to 0.1 sq cm with 2 decimal places
• Remember the state of the RTA "Show Distortion" button for the next startup
• Noise criteria values (calculated for the 1 octave RTA) are captured in the notes of saved RTA measurements

Bug fixes:

• Bug fix: Import sweep recordings timing signal search could fail for a captured file with a block of zero samples
• Bug fix: An error would occur if the alignment tool was used on measurements that did not have a phase response
• Bug fix: Try all available network interfaces when searching for 8c speakers
• Bug fix: Spectrogram slice plot did not reflect the normalise setting when the X axis was time
• Bug fix: T60 model trace could be drawn filled in some circumstances
• Bug fix: Ensure FS sine Vrms entries preserve more precision for small values, e.g mV or uV entries remain as mV or uV where appropriate
• Bug fix: Added a preference setting for the MacOS REW launcher to be excluded from the "Prefer Tabs" setting on Big Sur and so avoid a Java runtime bug that setting triggers

Changes in V5.20 RC5:

This build is mainly to address a few bugs, but there have also been some feature updates:

• Added an action in the EQ window target settings panel to create a measurement that has the current target shape, handy to use as a reference on the All SPL or RTA graphs
• A room curve will be added to the speaker driver target shapes if Add room curve is selected (previously the room curve was disabled for driver targets)
• Added an equaliser entry for the Emotiva RMC-1, it can also be used with the RMC-1L and XMC-2
• Only rewrite the FlexASIO.toml file if the "Close panel and update config file" button is used
• Removed the power saving hint from the macOS build
• Use 2 decimals for stepped sine dialog level controls
• Added buttons to reload the ASIO driver on the error dialogs shown if ASIO capture or ASIO replay fail to start

The bug fixes are as follows:

• Bug fix: Export all impulse responses as WAV ignored the min phase and equalised options
• Bug fix: With Java drivers the RTA title bar showed the wrong input channel
• Bug fix: Array index out of range could occur when aligning phase slopes with FDW applied
• Bug fix: on macOS the file dialog was always shown from the main REW window even if requested from another window
• Bug fix: Selecting Apply Defaults on the Waterfall graph for a measurement with no impulse response would cause an exception
• Bug fix: Moved ASIO driver loading back into event despatch thread to fix issues with the Komplete Audio 6 driver

Seems like only yesterday I was hoping RC3 would be the last V5.20 release candidate :/. Here we are with RC4 and two more bug fixes:

• Generating min phase response from Impulse graph could cause an exception
• Soundcard calibration could get stuck at "Calculating impulse response" during card response measurement

This is the 3rd and, ever optimistically, final V5.20 release candidate. Feature additions have been kept to a minimum, but there are some:

• There is now an export menu entry to export the RT60 data for all current measurements as text to individual files or to a combined file. Data is only exported if it has been generated (which happens when it is first viewed), a quick way to generate the RT60 data for all measurements is to view the Overlay RT60 graph.
• The limited range of the Impulse graph Offset t=0 dialog was causing some issues, so it has been changed to use a spinner that allows offsets of up to +/- 1 second to be entered, with 1 us changes when the spinner arrow buttons are used. To interactively adjust the t=0 offset click and drag the graph time axis.
• Processor loading has been reduced during measurement sweep playback
• 32-bit sample width will be used with Java drivers if the interface offers that. Only macOS and Linux are affected as the Windows Java runtime only offers 16-bit data (use ASIO for full sample resolution on Windows).

Several bugs have been fixed:

• RT60 calculation with time-reversed filtering could cause incorrect impulse envelope plot
• The T60M figure (RT60 Decay model T60 value) appeared twice in RT60 text exports
• Scope block processing could overrun at higher sample rates
• Array index out of range could occur when aligning phase slopes
• THD was not shown correctly when plotting stepped level results against input level
• Multi-channel average responses could be flat after one half or one quarter of the end frequency if the IR truncation preference retained more than 1.7 s of the IR