(June 1st 2018 : updated to version 1.2.1)
PaulXStretch – VST2/AU Plugin for extreme time stretching and other spectral processing of audio
This plugin is based on the PaulStretch algorithm. (Paul’s Extreme Time Stretch, originally developed by Nasca Octavian Paul.)
No manual or demonstration video exists at the moment, but some information follows :
The plugin is only suitable for radical transformation of sounds. It is not suitable at all for subtle time or pitch corrections and such. Ambient music and sound design are probably the most suitable use cases.
The plugin can either load an existing audio file from disk to process or it can capture audio from the plugin inputs. (The capture mode is started/stopped with the “Capture” parameter.)
The plugin will require the DAW audio engine to be running on the track it is inserted on. How that is achieved in various hosts is specific to those hosts. Some hosts will just immediately start running the plugin, in other hosts one may need to record/input monitor enable the track or something else needs to be done.
The plugin has lots of latency, that isn’t reported to the host as it likely can’t be properly compensated for, anyway.
Host automation does work, but in a laggy fashion and likely not completely deterministically. (Different play runs may produce different results.) This unfortunately can’t be helped much. The algorithm can only change the processing parameters between fully processed FFT blocks.
Lots of thanks to everyone who has donated so far! 🙂 👍
I have tested that the plugin loads and runs in :
Windows : Cockos Reaper 5, Ableton Live Lite 9, FL Studio, Plogue Bidule, JUCE plugin host (which may indicate it works in Tracktion/Waveform).
Mac Os : Cockos Reaper 5 (VST), Ableton Live Lite 9 (VST), Apple GarageBand (AU), Plogue Bidule (VST).
The plugin may well work in other hosts, but I am not able to test and debug with all of them.
There may be problems when doing offline rendering/export in Cubase. A workaround to get the plugin output into a final mix is to record the output of the plugin in real time by some means into an audio track and bypassing/removing the plugin before doing offline exports. (This may be a sensible workflow in other hosts too…)
v1.2.1 Windows 64 bit VST2, for Windows 7 or newer :
v1.2.1 Mac Os 64 bit VST2 and AudioUnit, for OS-X 10.9 or newer :
No Linux build yet, but will hopefully appear at some point. Support for 32 bit, Windows XP/Vista and Mac OS-X versions older than 10.9 is not in the plans. (People are of course free to attempt building for those platforms from source code but no attempt has been made in the source code to be compatible with those.)
Source code hosted at :
Change history :
-Added looping enabled parameter
-Added GUI button and parameter to rewind to beginning of selected play range
-Flush old stretched audio faster when source audio is changed
-Fix play range not being recalled properly when loading host project
-Changed “Octaves” module to “Ratios”. The Ratios module has more shifters than the previous Octaves module and allows changing the pitch ratios (and the shifters mix) in a separate tabbed page in the GUI.
-Spectral module enabled parameters changed to target named modules instead of chain slots
-Save and restore some additional settings
-Rebuilt with latest JUCE to fix parameter automation issue for example in Ableton Live
-Optimization in calculation of transformed free filter envelope
-Removed testing code that wasn’t supposed to be left in for 1.1.0
-Added free filter spectral processing module
*Yellow envelope line is the editable envelope, blue line is the envelope transformed with the free filter plugin parameters, used for the audio processing. (These are initially overlapped, tweak the free filter parameters to see them separated.)
*Click on envelope view to add point
*Alt-click on envelope point to remove point
*Alt-drag envelope line to change curvature
*Drag envelope line to move line up/down
*Right-click envelope view to show additional actions/options
-Made spectral module on/off states plugin parameters
-Fixed bug with the number of harmonics parameter not getting saved and recalled.
-Fixed bugs with calculations involving samplerate
-Added stretch processing bypass parameter (to play the original sound looped like it is passed into the stretcher)
-Show approximate stretched output duration in info label (only valid if the stretch amount is not automated in the host)
-Waveform selection can be moved by dragging the selection top area
-Smoothed playback with fades when changing waveform selection (doesn’t work ideally, fixing later…)
-Fixes for the waveform graphics disappearing unexpectedly (this probably still isn’t entirely fixed, though)
- Increased maximum number of input channels to 8
- Added zoom/scroll bar for waveform
- GUI performance improvement/bug fix during capture mode
- Shorter crossfade length when changing FFT size
-Control/Command click on waveform seeks (if click within active play range)
-Moved prebuffering amount menu to prebuffering meter (click to show)
-Added dummy parameter to tell the host the plugin state has changed when importing files etc.
(May not work properly for undo etc on all hosts.)
-Removed the factory presets as they are not really that useful
02-02-2018 1.0.0 preview 5
-Added buttons to enable/disable spectral processing modules
-Restored ability to set capture buffer length (via the settings menu)
-Seek to play range beginning when audio file imported
-No longer seeks to beginning of play range when changing FFT size
-Fixes to waveform display issues
-Double click on waveform selects whole waveform
-Double click on slider resets parameter to default value
-Show prebuffering ready amount graphically instead of text
-Highlight related parameters when spectral processing module clicked in the module chain
01-05-2018 1.0.0 preview 4
-Added reset parameters (except main volume and input pass through) command to settings menu
-Added option to settings menu to ignore loading imported audio file when recalling state
-Added support for dropping audio files to GUI (available functionality depends on host, many will simply
provide the file name of the source of audio file, so audio clip/event/item specific audio won’t be imported)
-Removed an unnecessary level of buffering (doesn’t reduce latency but should help a bit with CPU usage)
-Added About window
12-23-2017 1.0.0 preview 3
-Added parameter to set audio input capture buffer length (up to 120 seconds)
-Added parameter allow passing through audio from plugin input
-Inverted the number of harmonics parameter active range
-Added button to show settings menu
-Fix bug when offline rendering in host
-Allow setting background prebuffering amount in settings, including none.
(None is mostly useful in case the plugin doesn’t detect the host is offline rendering.
For real time playback none is likely only going to work with small FFT sizes.)
-Slightly better GUI layout but still mostly just 2 columns of sliders
-Remember last file import folder
-Added detection of invalid audio output sample values (infinities, NaN)
12-17-2017 1.0.0 preview 2b
-Fix buffer channel count issue, seems to make AU validation pass consistently
12-17-2017 1.0.0 preview 2
-Report only 2 input channels to host
-Fixes for using potentially invalid sample rates in internal initializations and calculations
-File import dialog allows importing file formats supported by JUCE, not just .wav
12-15-2017 1.0.0 preview 1
-Very early public release. Various issues present.