Akustyk / Blog: Recent posts

A very discerning user from Argentina (Thanks!) stubled upon a bug in the "Add signals..." procedure whereby if a sound file name started with a capital letter, the script would not run properly. This is due to the fact that the Praat scripting language does not allow variable names starting with an upper case character.

I fixed the bug in the new release (1.9.6) and now upper case characters are allowed.

Finally, the Watt and Fabricius vowel normalization method comes to Akustyk. It's come to my attention that people are interested in using this method in sociophonetic analysis of vowel space. So here it is.

Create global synthesis...

The PSOLA and LPC-based synthesis functions have been among Akustyk's most popular features. In version 1.9.2, I added another module, Create global synthesis.... The module adds F4, F5, B4, and B5, and allows the user to check formant tracks in a Praat Formant Grid object to look for possible "jumps" or "drop-outs." This feature was designed by Anne van Leeuwen (thanks!). It will be particularly useful to researchers who need to control five formants and five bandwidths independently. The user interface and general principle is very similar to that of Create basic synthesis... and you can find out more by viewing this video tutorial.

Installers

Since the early days, Akustyk has had a Windows installer (wizard) and a Mac OS X installer. Because I don't have a Mac machine, I quit making the Mac installer and offered source code with instructions on how to install. When OS X Lion came out, the user's Library became harder to find, thus making installation more difficult. Besides, Akustyk was not conforming to Praat's plug-in specification.

Praat plug-in architecture

Some time ago (not sure exactly when) Praat introduced a plug-in API. It simplified and standardized the way in which plug-ins could be integrated into the Praat user interface and run scripts and programs from within Praat windows. For the sake of backwards compatibility and due to my own laziness, I did not adopt the model for Akustyk. Until now...

One installer

I have finally gotten rid of all non-native code and wrote a simple Praat script that installs Akustyk on Windows, Linux, and Macintosh. I tried to simplify the installation (and uninstallation) process.

Still too hard?

The installation might still be a little too hard. One needs to download the archive, uncompress it to the Desktop, open the install script and run it. I need to think of simplifying the process. Perhaps I should use a self-extracting archive? I really don't want to build binary installers any more. I hope this new installation process works better than the old one!

Save as PNG file...

EPS file format

Praat offers a truly wonderful plotting functionality. One of the things that make it special is the ability to save images as EPS files. The EPS file format is a so-called vector image format and it has been the gold standard for high-quality print publishing for years. However, the format is not as ubiquitous in the Web and desktop environments where bitmap image formats dominate. Therefore, if you want to include your Praat (or Akustyk) image in a PowerPoint presentation, share it with colleagues via email, insert it in a web page, etc., you have to convert the EPS image into a more commonly used raster (bitmap) format.

PNG file format

The Portable Network Graphics file format is ubiquitous throughout the desktop and Web environments. Due to its remarkably good handling of text, grayscale, and color, it is particularly suitable for scientific images of all kinds. It is, in my opinion, the best file format to display spectrograms. It supports transparency and has a large number of other useful options.

ImageMagick

I wanted to use an open-source image processing library for this purpose. Without hesitation, I picked ImageMagick. If you're familiar with SOX, ImageMagick is the image processing equivalent of the audio processing SOX. It works on many platforms, including Linux, Macintosh, and Windows. In order to use this new function (Save as PNG file...), you're going to have to install ImageMagick first, but it is a fast and painless installation. It's worth having the program on your system in any case.

How to use it

The function is available from the Praat Picture Window inside the "File" menu. When you click on the link, you're going to get a small dialog box (see below) with a few options to choose from. I recommend that you start with the default settings: you'll end up with an EPS image and its PNG equivalent, on your Desktop. The "Simple" mode includes four options, ranging in quality from Web to Print.

Advanced mode

If you need more control, you'll need to use the advanced mode. Here, you set your own resolution, choose transparency (or flattened image), set precise dimension for "fit" (fit in a box), "fill" (fill a box), or a percentage-based scaling. For optimal results, I recommend setting the same value for the resolution and re-size. That way, you'll end up with the best combination of resolution and size in pixels.

Send in feature requests and bug reports

If you think you need additional functionality, or you've found bugs in the program, do not hesitate to email me.You can download Akustyk 1.9.1 Beta here.... read more

Support for Catalan and Cantonese

Thanks to the contributions from Akustyk users, we now have added support for Catalan and Cantonese. To clarify, yes, you can set up custom systems for full control over the inventory, but for those users who prefer the convenience of having pre-configured vowel systems available, I will keep adding systems to Akustyk. This comes in handy particularly in the classroom. It would be much more difficult for each student to have to set up a custom inventory for a class project. Instead, they can take advantage of the systems already available.

Setting up preferences

Before you can start analyzing speech, you need to set up session preferences. You need to let Akustyk know which vowel system you are planning to use. You also need to tell Akustyk a bunch of other things, but, fortunately, all of them are optional and the defaults should be able to suit most sociophonetic analysis.... read more

Add custom sound system

Previously, if one wanted to use Akustyk's automated speech analysis and plotting tools, one had to use one of the pre-configured vowel systems. This worked for a lot of people, but, every now and then, I got emails from linguists who wanted to analyze a sound system not currently supported by Akustyk. I'd often simply add that system myself, especially if more people had asked for it. Now, it is no longer necessary.

Wizards

I've thought about how to implement the new feature for quite some time. I wanted the process to be very simple, done by means of a form window, rather than requiring scripting skills on the part of the user. I first considered writing a separate application in C, but scrapped the idea due to the amount of work required to make it work cross-platform. A much more realistic choice would have been PHP, in a browser window. But then Praat implemented a more advance "pause" function. This enabled me to create a wizard-like interface.... read more