flancast90 / speech-to-text-in-tw5 Goto Github PK

Just need a one-liner solution to get the language of TW (not plugin), TW itself. I am going to implement an option to change the plugin language when TW language changes.

@flancast90 I would like to use the speech to text plug in in my wiki. But I haven't seen any updates for the plug in in the recent past. Is the plug in development abandoned?

As the title implies, we should make it clear that this plugin works best on Chrome/Chromium and under NodeJs

What makes it work best under NodeJs is the granting of permissions for microphone access
Chrome/Chromium can permanently allow microphone access to websites using the https protocol but not to local files ( file:/// )

Hi all,

This is just a simple post to explain the documentation I will add to the README.md. I plan to include

• Project Description
• Install
  a. Drag-and-Drop
  b. Dev setup for Node.js
• Use Cases
  a. basic "Get Started" for new tiddler with transcript
  b. transcript in existing tiddler (when applicable)
  c. verbal commands (custom, too, when applicable)
• License (MIT)

anything else anyone can think of?

As requested, I have opened a new discussion for the implementation of the feedback received. I intend to elaborate more on the command keywords, as well.

We could add a config which, if set to yes restarts the recognition automatically when it would end by timeout. Then it would only stop if it's stopped by hand via button-click or keyboard shortcut or via voice-command stop listening

@BurningTreeC ,

Just wanted to let you know that I have been working on TW-Sounds, and analysis of sounds and similar-sounding words to integrate with this plugin! You can see what I have done so far at https://github.com/flancast90/TW-Sounds. It is not quite ready, but I am working on the part for finding rhymes and near-rhymes without API. As you can imagine, the natural language processing and maths involved are a bit much!

I open this issue as a reminder that we want to create a text-editor toolbar button that allows inserting the recorded text into the current tiddlers' text field

For that we need some brainstorming and we need to collect information about the tiddlywiki internals so that we know how we can realize this

The getTranscriptCommands function executes commands
There's a logical error somewhere which causes commands being executed twice

I need help in solving this problem

• We can set recognition.continuous = true to extend the recognition of the speech
• We then can concatenate results of the speech recognition

The limitation is, that a voice-command cannot be the same as any of the defined OK-keywords
But that's a limitation which should be fine if we document it

There's a problem:

• A transcript can finish with Ok Wiki and the next transcript can start with the actual command
• We should overthink how we handle the command-recognition
• We should concatenate all transcripts together and then check for the Ok Wiki and the following command
• Having found such a command-chain, we remove it from the full transcript
• Then we process the command

Hi @BurningTreeC,

Since neither of us have been real active lately in this repo, I've decided to start work on v1.0.7/8, utilising TW-Sounds. I'll add the functions and calls, and if you want, you can look over the code and make it more "tw-like".

In the current master we can create tiddlers with tag $:/tags/VoiceCommand and the field voice-command which contains the command that should trigger the action. The text field should contain the actions that are then triggered.

There's also the possibility to define your own custom keywords additionally to Ok and Wiki using the tiddler $:/config/speech-to-text/keywords and its fields ok-keywords and wiki-keywords

The field ok-keywords could then contain "Pasta pasta" and the field wiki-keywords could contain "Maccheroni macaroni" and then Pasta macaroni switch language to german would work

Hi @flancast90 , I'm writing this first issue to get an overview how we want to start this little project :)

@BurningTreeC:

I just wanted to thank you for all the help you've been so far in how to make a TW-plugin, and wanted to announce that I've made my first solo plugin: text-to-speech. The plugin is really quite simple, and I got the idea from TiddlyTweeter's question about this functionality. I felt that it wouldn't fit this plugin well (a bit off-topic for this plugin), so I created a minimal implementation of something here.

Please add Persian language

We should now test test test so that we can release v1.0.5

• Is language switching stable (by all 3 methods)?
• Do user-defined voice commands work consistently?
• Do user-defined keywords work consistently?