Simplest Method for Speech To Text messages?
Hello everyone!
I'm trying to create a max patch that will take in audio from a microphone, convert it from speech to text (STT), then put that text into a message box for later use. I've been trying out different methods to do this for the past few weeks but haven't gotten anything to work in Max. op.recognize doesn't seem to work with Maxmsp 7, I found a patch that used the Google API by Tom Zicarelli, but that doesn't seem to work properly. I'm not very well versed in working in Max on this level, so I don't even know which things that don't work are closer to working than others. I'd really appreciate some advice on how to go about this.
Thank you.
Ever get it to work?
A few weeks ago I prepared an example using Google Chrome STT engine and a simple bridge (based on websockets) to transfer recognised data to Max over Node.js.
I posted this example with some instruction here: https://cycling74.com/projects/project35-oprecognize (scroll the topic for my post with ZIP archive attached ).
I ended up basing my project on this project: http://florianschulz.info/stt/
I simply modified the processing sketch to use OSC to send the STT results to Max via the udpRecieve object. It's a little convoluted, but the timeframe of the project meant I needed to prioritise quicker progress vs long term viability.
It does require Processing, but what Yaniki just posted is very very similar and would probably be more stable in the long run.
I think both methods have strong and weak points, depends on the context. Moreover my intuition tells me to check whether we can use Node.js to STT.
I've got speech recognition to work on a browser (with annyang.js) and am also looking into how this could work on Max MSP.
Saw this video on YouTube:
Anybody has any idea how he did it?
He did it using the built-in dictation of MacOS. It has to be enabled through the settings, and then any text fields can be dictated to.
System Settings -> Keyboard -> Dictation -> On
@TANSY X
How can i change the language?