Posted on

IPA Keyboard Layout

Well, it turns out, there is no such thing, per se, as an IPA Keyboard Layout, at least not in the sense that there are keyboard layouts for various languages and layout styles (e.g., English, Dvorak, etc.). This seems to me to be a tremendous oversight, though it obviously came about because someone thought supporting the entire Unicode space for the IPA was a great idea, and the only idea. There are two things needed to have an IPA Keyboard Layout that would be functional for someone working in one or a few select languages: - A Keyboard Layout File, such as discussed here for X11 support (Linux) - One or more fonts that provide the support needed -- which includes a wide variety of unicode symbols plus the specific ipa unicode extension block. And preferrably fonts with multi-lingual support so that a mixture of IPA and one or more languages would by typographically elegant, or at least not jarringly unaesthetic.

IPA Character Support for a Given Language

The first step is to get a useful/functional/popular mapping of a language to IPA characters. For a language with diverse dialects, some standard form needs to be determined. Along with this is the likelihood of supporting two languages. For example, if the target language is Thai, the supporting/documenting language is likely English or another foreign language. Both Thai and English IPA character spaces need to be determined, and together they should map out the shared space, in a Venn-like diagram.

Determine Key Mapping Desired for Charcter Space

A simplistic approach would be taking the 26 characters in English and mapping those keys (lowercase and uppercase) to obvious matches, and expanding into punctuation keys as needed. Requirements for dead keys and multi-keystrokes in general might be avoidable. The main approach should be to reuse as much of the current set of wheels available rather than re-inventing one's own.

Build Keycap File and Print Keycaps

It seems straightforward to have keycaps that would support two languages and IPA. This would provide a nice intermediary, additional script which could support both of the other two languages. For some languages which maintain a large portion of the alphabetic character space in English, a third (fourth) script might be able to be acommodated, such as: English, IPA, Thai, and Vietnamese; perhaps even English, Indonesia, IPA, Thai, and Vietnamese.

Posted on

DeepSpeech – Machine Learning

DeepSpeech Speech Recognition Machine Learning

These are notes to the project, which seem to me worth pursuing. Having recently seen a number of AWS re:invent videos on Vision and Language Machine Learning tools at Amazon, I have ML-envy. Time to start a project, but while I wait for the Amazon Transcribe and Amazon Translate to become available, the recently released Mozilla DeepSpeech project looks interesting. More resources as follows: - Mozilla DeepSpeech - Github - OSX 10.11 issues - Taking DeepSpeech for a spin - Machine Learning at Mozilla - Kaggle resources on ML (also specifically DeepSpeech)