Considerate la vostra semenza:
fatti non foste a viver come bruti,
ma per seguir virtute e canoscenza.
– Dante Alighieri, Inferno, Canto XXVI
I’m proud to present Parlatype 3.0 together with Parlatype LibreOffice Extension 3.0. This release brought mostly changes in Automatic Speech Recognition (ASR) and a lot of work “behind the scenes”.
Parlatype got a refreshed icon, kindly provided by Tobias Bernard.
Added support for Mozilla DeepSpeech.
ASR works now only with the LibreOffice Parlatype Extension. Other means of text output were removed. On the other side the parameter
--with-asris not necessary anymore.
There is a new ASR configuration system based on single language models. A configuration points to a third party download location and holds all the information necessary to use that language model. Those configurations can be easily changed or added.
Although I’m quite excited about it, this release is actually not very exciting – but please don’t tell anyone. It’s mostly work behind the scenes, invisible to the user:
Whoever needs to build Parlatype from source, please have a look at the NEWS file in the root directory which tells you about changed dependencies.
If you build from git: The development branch is now called “main”, branch “master” has a README.md how to switch the branch and will be deleted in the future (Parlatype and LibreOffice Extension).
Rewrite the waveviewer widget to ease future switching to GTK4. There is no visible difference and there are no known regressions yet. The code base saw numerous other changes in preparation for GTK4.
Modularize backend code for better maintainability.
Make GStreamer plugins private. Some users reported that Parlatype’s plugins were not registered with GStreamer. These problems (which I could not reproduce) should be solved now.
Fixed memory leaks.
Update flatpak manifest: Users can build Parlatype with Mozilla DeepSpeech.
Thanks to Launchpad translators there are updated translations for German, Italian, Hebrew and Polish.
For the LibreOffice Extension the Italian translation was updated.
I’m pleased that a couple of distributions are shipping Parlatype. They will pick up this version and you should be able to install Parlatype with your distro’s tools.
Things look a bit different when it comes to Automatic Speech Recognition (ASR), especially Mozilla DeepSpeech. ASR libraries are hardly packaged. I hope to make Parlatype available with DeepSpeech support on Flathub in the short term. Alternatively you would have to install it from source.
As far as Windows is concerned: Unlike the previous version, this release doesn’t build on Windows.
Personal note and prospect
This release took much longer than planned. You might have heard of a pandemic going on – it brought some changes to my work life. Also the code base has grown over time. Being developer, maintainer, packager started as fun, became (unpaid) work and turned into a burden. I have to admit that I was exhausted for some time and struggled with motivation.
I hope to continue this work. I will have to manage the project a bit better and I even have a faint hope to find others who would like to contribute.
The future direction of the project is amibigious: Parlatype started as a tool for manual transcription, then my focus turned to Automatic Speech Recognition which I find more interesting – but also much more complicated.
I’d like to thank Fabian for being the first person who donated via PayPal to Parlatype. Thank you, Fabian! I invite you, dear reader, to become the second person. Just go to to the donation page.
Financial transparency report: Since the last release (actually since the first release in 2016) I have received the amount of 10 € for my work and the expenses I have for website hosting.
Signing key has changed
This release is signed with a new key. To get the new key see how to verify the signature.
Download source code
The links below point you to the source code and an installable extension (Parlatype.oxt):