PXToolkit Beta Available Now!

[Lady Gaia]

That works great, thanks.

With the rename, what would be nice is if you could take in a list of original filenames and batch process these using px-renumber.

I've seen enough feedback from some who aren't comfortable with the command-line that I'll probably opt to rework it with a friendly UI eventually before adding anything to the current functionality.  It's not an immediate priority, but the name "PXToolkit" was chosen to cover a range of possible features beyond importing samples.  Archive renumbering is certainly on the list of possibilities.

[StephenJamesBennett]

Thanks again—renaming 20 files nearly killed me

I'll take a little more care the next session ...

Stephen

[Perilous]

I think I didn't a bad thing. What does this error mean?



(Windows 10)

[Lady Gaia]

I think I did a bad thing. What does this error mean?

It either means that the file 21.wav isn’t a well-formed .wav file, or that PXToolkit doesn’t understand some infrequently used aspect of the .wav file specification.  If you can upload 21.wav somewhere like DropBox and PM me a link to it I’ll dig into the file to figure out which is the case.

(If I had to guess, I’d say the file is longer than necessary and just has a bunch of zero byte padding at the end.  The duplicate “chunk” identifier is probably 00 00 00 00.  I realize this is more me talking to myself than anything.  Users shouldn’t have to know anything about the file format, but the error messages are there to help diagnose situations like this.)

[Gerry Havinga]

I've been trying to figure out what best tools I can use on Windoze or Linux to manipulate the wav file headers/trailers to insert the correct pitch data. Having the root pitch of the sample inside the meta-data will make life a lot easier importing the sample. Only filename conventions do not seem to work very well.

Does anyone have any tips here, what best tools to use (preferably Linux command line for batch processing)?

[Lady Gaia]

I leave the Linux and Windows tools question open to others, since macOS is my home turf.

Regarding the inference of root pitch from filename, though, that should be pretty reliable.  I'll only use it if there isn't authoritative metadata in the .wav, but I'll recognize a pretty wide range of formats.  What I'm looking for is:

* A delimiter (can be a space, dash, period, underscore, tilde, plus sign, or any kind of opening parenthesis/bracket/brace)
* Followed by a letter from A to G (lower case is also acceptable)
* Optionally followed by # or b
* Followed by an octave integer (anything from -2 to 8 )
* Followed by another delimiter (like the opening delimiter but with closing bracket/brace/parenthesis)

So it should be pretty flexible.  The one limitations that leap to mind are that you can't embed extra or other unrelated characters in the middle of the note name, you can't use a + in front of the octave, and it won't recognize H as a valid note.  I couldn't come up with a reasonable heuristic for what B should mean that isn't ambiguous once you recognize H as a possibility.

[Shaw]

This is a quality piece of work.  Easy to use, solid, does what you need...  many thanks from many people! 
DSI should license this from you and just ship it with every PX.  Well done, Madame, well done!

[Gerry Havinga]

I leave the Linux and Windows tools question open to others, since macOS is my home turf.

Regarding the inference of root pitch from filename, though, that should be pretty reliable.  I'll only use it if there isn't authoritative metadata in the .wav, but I'll recognize a pretty wide range of formats.  What I'm looking for is:

* A delimiter (can be a space, dash, period, underscore, tilde, plus sign, or any kind of opening parenthesis/bracket/brace)
* Followed by a letter from A to G (lower case is also acceptable)
* Optionally followed by # or b
* Followed by an octave integer (anything from -2 to 8 )
* Followed by another delimiter (like the opening delimiter but with closing bracket/brace/parenthesis)

So it should be pretty flexible.  The one limitations that leap to mind are that you can't embed extra or other unrelated characters in the middle of the note name, you can't use a + in front of the octave, and it won't recognize H as a valid note.  I couldn't come up with a reasonable heuristic for what B should mean that isn't ambiguous once you recognize H as a possibility.

Thanks Lady Gaia. The issue might be because of the first delimiter, there is none before the note name. By the way, neither your PXToolkit nor the 8Dio tool work on these filenames. They both behave the same and assign wav files in alphabetical order to the keys. Filenames are:
A#2.wav
A2.wav
A#3.wav
A3.wav
A#4.wav
A4.wav
B2.wav
B3.wav

and so on.

Perhaps i should try to put a character + delimiter (underscore) in front of the note, something like:

fred_A#2.wav

Obviously I can drag the filenames myself, but being a lazy person, I'd prefer to script it.

[Lady Gaia]

I've heard from a few folks who were using earlier versions of PXToolkit without being aware of updates, so as of last night's 0.9.0 release, PXToolkit will automatically check and provide a reminder when there's a newer version available.
 

Thanks Lady Gaia. The issue might be because of the first delimiter, there is none before the note name.

That's not a naming scheme I had considered as I assumed something would always precede the note name, but it's easy enough to fix.  As of 0.9.1, which is already available, the opening delimiter will no longer be necessary if the first character of the filename is the start of the note name (and the closing delimiter could already be omitted if the note name is immediately followed by the .wav extension.)

It meant wading back into regular expression land but only briefly, so no lasting harm was done.  Would you believe "^(?:.*[-. ~+_({<\[])?([a-gA-G][#b]?(?:-[12]|[0-8]))(?:[-. ~+_)}>\]].*)?$" describes files that match the range of naming schemes I recognize?  Yikes.

Version 0.9.1 also attempts to ignore cases where there's unnecessary padding at the end of a .wav file.  This should take care of Perilous's case above, but I'd appreciate confirmation from anyone who saw similar symptoms.

[Gerry Havinga]

I've heard from a few folks who were using earlier versions of PXToolkit without being aware of updates, so as of last night's 0.9.0 release, PXToolkit will automatically check and provide a reminder when there's a newer version available.
 

Thanks Lady Gaia. The issue might be because of the first delimiter, there is none before the note name.

That's not a naming scheme I had considered as I assumed something would always precede the note name, but it's easy enough to fix.  As of 0.9.1, which is already available, the opening delimiter will no longer be necessary if the first character of the filename is the start of the note name (and the closing delimiter could already be omitted if the note name is immediately followed by the .wav extension.)

It meant wading back into regular expression land but only briefly, so no lasting harm was done.  Would you believe "^(?:.*[-. ~+_({<\[])?([a-gA-G][#b]?(?:-[12]|[0-8]))(?:[-. ~+_)}>\]].*)?$" describes files that match the range of naming schemes I recognize?  Yikes.

Version 0.9.1 also attempts to ignore cases where there's unnecessary padding at the end of a .wav file.  This should take care of Perilous's case above, but I'd appreciate confirmation from anyone who saw similar symptoms.

Ha ha very funny, I love regular expressions when I can get them to work for me...... I've done a lot of sysadmin style automation using good ol' Perl in my days. Switching more to Python and Ansible (not really a scripting language, but still very cool) nowadays.

I figured putting the filenames in a sample voice named directory structure, was enough for me to recognize what I was working on. Also makes it quite easy to archive the samples using tar or 7zip and sent them around my computer systems.

Still it would be great to be able to hack the audio meta data. I have downloaded a trial version of Sample Robot, but at the moment I don't have the time to get a degree in  astrophysics to understand even the interface.

I will be testing your new version later tonight. Don't require a degree for that ;-). Thanks for keeping the interface simple and clear, but complex enough to quickly do the most important task of mapping the samples.

[Perilous]

It either means that the file 21.wav isn’t a well-formed .wav file, or that PXToolkit doesn’t understand some infrequently used aspect of the .wav file specification.  If you can upload 21.wav somewhere like DropBox and PM me a link to it I’ll dig into the file to figure out which is the case.

Thanks, LG! I hope to have the files uploaded soon. I am using an old outdated wave editor to trim a large wav. into samples. It works with the 8dio utility but I still get the same error even with the latest version of PXToolKit. I would send one of those newfangled PMs but it appears I must first brush up on my DSI/Sequential history.

[LoboLives]

I really think the loop parameter in the Prophet X should just be available in the mapping software and you can create loop points and specify what samples you want to be automatically looped by default.

[Lady Gaia]

Thanks, LG! I hope to have the files uploaded soon. I am using an old outdated wave editor to trim a large wav. into samples. It works with the 8dio utility but I still get the same error even with the latest version of PXToolKit.

Hmmm.  I'll be happy to help you get to the bottom of this.  I'll PM you a link to a shared folder on DropBox where you can upload the problematic file (unfortunately you will need to create an account there to use it, but they're free.)

Ha ha very funny, I love regular expressions when I can get them to work for me...... I've done a lot of sysadmin style automation using good ol' Perl in my days. Switching more to Python and Ansible (not really a scripting language, but still very cool) nowadays.

Most of my work is in native languages, but for PXToolkit I wanted something that would give me a quick way to reach both macOS and Windows users.  Would you believe the whole thing is written in TypeScript?

Still it would be great to be able to hack the audio meta data. I have downloaded a trial version of Sample Robot, but at the moment I don't have the time to get a degree in astrophysics to understand even the interface.

SampleRobot definitely has that classic engineer-designed user interface but it's quite capable once you get past the quirky character.  I find it invaluable for trimming, looping, and normalizing samples.

I will be testing your new version later tonight. Don't require a degree for that ;-). Thanks for keeping the interface simple and clear, but complex enough to quickly do the most important task of mapping the samples.

I'm glad it's hitting the right sweet spot for you.  The art of designing something usable is a tricky one and can be rather subjective, so I'm never sure until someone else tries to use one of my designs.  I'm still struggling to figure out how best to expose the one esoteric feature the Prophet X enables that I haven't incorporated yet, precisely because it's not something I think needs to be prominent: you can specify on a per-sample basis how stereo samples should be collapsed to mono when it's required - either by summing both channels or just using the left or right channel on their own.  Right now PXToolkit just marks everything as collapsing by summing the two channels.

I'll keep tinkering with ideas, because I really don't want to clutter things in the process for a feature many won't need to worry about.

I really think the loop parameter in the Prophet X should just be available in the mapping software and you can create loop points and specify what samples you want to be automatically looped by default.

Samples with loops already defined are already mapped correctly.  Defining new loop points requires a very different user interface as you really need to see the waveform to make decent sample-accurate decisions.  I've debated whether to tackle that project eventually or not but it definitely isn't my immediate goal as it would increase the scope significantly and start looking more like a viable commercial product.  It makes sense to treat sample format and rate conversion, trimming, normalizing, looping, etc. as a whole family of related problems.  For now my personal tool of choice in that space is SampleRobot.

[Gerry Havinga]

Thanks Lady Gaia, you nailed the regular expression on the head! It works perfectly now, samples are properly assigned, even with the quirky filenames I have used. Good job.

[Perilous]

Hmmm.  I'll be happy to help you get to the bottom of this.  I'll PM you a link to a shared folder on DropBox where you can upload the problematic file (unfortunately you will need to create an account there to use it, but they're free.)

Thanks. It's not one problematic file, it's each and every one. DropBox is also problematic.
I can try this: https://drive.google.com/open?id=157R1BVNKt2WhKxXNfCBuUJg98flddGui
I'm using an old wave editor to chop up the long .wav into its own separate file for each note.
This old program is likely the culprit. The only output settings I have is PCM Uncompressed, and I think it's supposed to be 44100 Hz, 24 Bits, Stereo, if I don't have my head screwed on backwards (which has been the case all week). The master file that I'm chopping from actually imports perfectly fine.

[Lady Gaia]

Version 0.9.5 of PXToolkit is now available.  It includes a number of small improvements, including compatibility with technically invalid .wav files along the lines of the example @Perilous kindly provided (though as always they'll need to be 48kHz 16-bit for use with the Prophet X.)

This release also adds support for the only Prophet X user instrument feature I'm aware of that was not previously supported.  The included README describes the new Advanced Sample Features context menu as follows:

Several esoteric features relating to samples are now available on a context menu. Right-click (or control-click on macOS) on either an individual sample to modify an individual sample, or anywhere in the canvas to affect all samples in currently selected regions. The context menu offers the following advanced options:

If one or more selected samples are looped, the menu will show the item "Use Sample Loop." Toggling this allows the loop described in .wav metadata to be ignored when the instrument is exported.

If one or more selected samples are in stereo, the menu will show four options for controlling how sound is collapsed to mono in 16-voice mode. Four options are available: summing both left and right channels, using only the left channel, only the right channel, or defaulting to whatever mode has been selected for the instrument as a whole, as described below.

The final section of the menu allows the default mono collapse mode to be selected. These options will be displayed no matter what is currently selected, and affect all stereo samples that do not have an explicit mode selected.

A checkmark will appear beside the currently selected mode in each case. When two or more samples are selected it's possible that multiple modes are active. This is reflected by three dots beside the modes that aren't uniformly applied. Selecting a new mode will cause all selected sample to change to the same value.

The sample detail list will reflect new states as follows:

* Samples with loops that have been disabled will display "Loop Disabled" instead of "Looped"
* Stereo samples that aren't using the instrument-wide default mode for collapsing to mono will display "(( Stereo ))", "(( Stereo", or "Stereo ))" to reflect using both channels, just the left channel, or just the right channel respectively.

[Lady Gaia]

Thanks. It's not one problematic file, it's each and every one. DropBox is also problematic.
I can try this: https://drive.google.com/open?id=157R1BVNKt2WhKxXNfCBuUJg98flddGui
I'm using an old wave editor to chop up the long .wav into its own separate file for each note.
This old program is likely the culprit.

That was very helpful.  Thank you!  The file is indeed technically invalid, but I've included changes that will allow files with this particular flaw to be recognized by PXToolkit.

The only output settings I have is PCM Uncompressed, and I think it's supposed to be 44100 Hz, 24 Bits, Stereo, if I don't have my head screwed on backwards (which has been the case all week). The master file that I'm chopping from actually imports perfectly fine.

This particular file, however, is 44.1kHz 24-bit as you noted, and the Prophet X only supports 48kHz 16-bit samples, either in mono or stereo.  PCM Uncompressed is also required, so you're okay on that front.  With a tool of a vintage that produces this particular kind of quirky .wav file, you're almost certainly better off not having it do any sample rate conversion in any case, and even bit-rate changes may not be optimal.  Since the full file imports without issues, the original must already be 48kHz 16-bit, so just stick with that format and you should be good to go!

[Proton]

Thank you again, Lady Gaia ... fantastic work. Very appreciated.

[Perilous]

That was very helpful.  Thank you!  The file is indeed technically invalid, but I've included changes that will allow files with this particular flaw to be recognized by PXToolkit.

That's really cool. It works now! Thanks a lot. I was going to ask what do folks use for slicing and dicing when I remembered I already have Audacity installed. Derp. I guess my first reaction was to use the quirky wave editor because the chopping process is quicker. Thanks again for your marvelous support.

[armenjc]

Just a heads up that beta version 0.9.8 of PX Toolkit is up on ThinkerSnacks - It still reads "beta 0.9.7" under the download button, but the file you download will be 0.9.8.  I don't know all the specifics of the newest version other than Lady Gaia went to the trouble to update the app so we can now drag 44.1k files into the GUI! Hot damn!