text recognition

[watsonlisp]

Bing AI spit this out, apparently python.
Is there and easy way to use it for speech recognition?
maybe someone with a compiler could make me an exe?



import speech_recognition as sr
import win32com.client

# Initialize the recognizer
r = sr.Recognizer()

# Initialize AutoCAD
acad = win32com.client.Dispatch("AutoCAD.Application")

# Listen for speech
with sr.Microphone() as source:
print("Speak:")
audio = r.listen(source)

try:
# Recognize speech using a hypothetical API
text = r.recognize_google(audio)
print(f"You said: {text}")

# Send the recognized text as a command to AutoCAD
acad.ActiveDocument.SendCommand(text + "\n")

except sr.UnknownValueError:
print("Could not understand audio")
except sr.RequestError as e:
print(f"Could not request results; {e}")

[daniel-cadext]

Too much fun!

Using the Python wrappers at
https://github.com/CEXT-Dan/PyRx

import PyRx as Rx
import PyGe as Ge
import PyGi as Gi
import PyDb as Db
import PyAp as Ap
import PyEd as Ed
import traceback

import speech_recognition as sr

def PyRxCmd_doit():
try:
r = sr.Recognizer()
with sr.Microphone() as source:
print("Say something!")
audio = r.listen(source)
try:
Ap.DocManager().sendStringToExecute(Ap.curDoc(), r.recognize_sphinx(audio)+"\n")
print("Sphinx thinks you said " + r.recognize_sphinx(audio))
except sr.UnknownValueError:
print("Sphinx could not understand audio")
except sr.RequestError as e:
print("Sphinx error; {0}".format(e))

except Exception as err:
traceback.print_exception(err)

but it cant understand me, maybe my microphone

Command: PYLOAD
Success module _DOIT is loaded:
Command: DOIT
Say something!
Sphinx thinks you said nine months
Command: DOIT
Say something!
Sphinx thinks you said it drop a line
Command: PYLOAD
Module _DOIT Already loaded, use pyreload
Command: PYRELOAD
Success module _DOIT is reloaded:
Command: DOIT
Say something!
Sphinx thinks you said draw ally and that
Command: draw Unknown command "DRAW". Press F1 for help.
Command: ally Unknown command "ALLY". Press F1 for help.
Command: and Unknown command "AND". Press F1 for help.
Command: that
Unknown command "THAT". Press F1 for help.
Command: DOIT
Say something!
Sphinx thinks you said catherine so-so
Command: catherine Unknown command "CATHERINE". Press F1 for help.
Command: so-so
Unknown command "SO-SO". Press F1 for help.
Command: DOIT
Say something!
Sphinx thinks you said drop caught lying i'm not fun but what can
Command: drop Unknown command "DROP". Press F1 for help.
Command: caught Unknown command "CAUGHT". Press F1 for help.
Command: lying Unknown command "LYING". Press F1 for help.
Command: i'm Unknown command "I'M". Press F1 for help.
Command: not Unknown command "NOT". Press F1 for help.
Command: fun Unknown command "FUN". Press F1 for help.
Command: but Unknown command "BUT". Press F1 for help.
Command: what Unknown command "WHAT". Press F1 for help.
Command: can
Unknown command "CAN". Press F1 for help.
Command: DOIT
Say something!
Sphinx thinks you said that i may know who
Command: that Unknown command "THAT". Press F1 for help.
Command: i INSERT
*Insert a block from the Blocks palette
Command: may Unknown command "MAY". Press F1 for help.
Command: know Unknown command "KNOW". Press F1 for help.
Command: who
WHOHAS
Command: *Cancel*
Command: *Cancel*
Command: *Cancel*
Command: DOIT
Say something!
Sphinx thinks you said and sure he
Command: and Unknown command "AND". Press F1 for help.
Command: sure Unknown command "SURE". Press F1 for help.
Command: he
HATCHEDIT
Select hatch object: *Cancel*
Command: DOIT
Say something!
Sphinx thinks you said and sure
Command: and Unknown command "AND". Press F1 for help.
Command: sure
Unknown command "SURE". Press F1 for help.
Command: DOIT
Say something!
Sphinx thinks you said has she added that uh huh
Command: has WHOHAS
Command: she Unknown command "SHE". Press F1 for help.
Command: added Unknown command "ADDED". Press F1 for help.
Command: that Unknown command "THAT". Press F1 for help.
Command: uh Unknown command "UH". Press F1 for help.
Command: huh

[daniel-cadext]

Sorry, I don't know how to edit my posts


github CEXT-Dan/PyRx Python wrappers for ARX

import PyRx as Rx
import PyGe as Ge
import PyGi as Gi
import PyDb as Db
import PyAp as Ap
import PyEd as Ed
import traceback

import speech_recognition as sr

def PyRxCmd_doit():
    try:
        r = sr.Recognizer()
        with sr.Microphone() as source:
            print("Say something!")
            audio = r.listen(source)
        try:
            Ap.DocManager().sendStringToExecute(Ap.curDoc(), r.recognize_sphinx(audio)+"\n")
            print("Sphinx thinks you said " + r.recognize_sphinx(audio))
        except sr.UnknownValueError:
            print("Sphinx could not understand audio")
        except sr.RequestError as e:
            print("Sphinx error; {0}".format(e))
       
    except Exception as err:
        traceback.print_exception(err)

[daniel-cadext]

BTY, I’ve wrapped a good chunk of ObjectARX for Python (still a work in progress)
The source and compiled modules are on github.

Played around with AI and Eye detection. Still wondering if I can train an AI to detect CAD geometry.
But, If your drawings have cats in them, I should be able to locate them lol

[watsonlisp]

thanks daniel, can you show me how to use it?

[Ed Jobe]

Sorry, I don't know how to edit my posts

Hi Daniel
There is a link in the lower right of your post, next to Go Advanced. Also, you can edit your username to get rid of the numbers.

[daniel-cadext]

Hi,
Python for AutoCAD, search GitHub for CEXT-Dan/PyRx, there’s a release folder with installers for AutoCAD2021-2024. I’ve made wrappers for both ActiveX and ObjectARX. There should be enough samples to get you started, its still beta, but quite usable

As far as the speech_recognition module, I only tried it once, so I’m certainly not in a good position to give advice. Except maybe a couple thoughts

As you can see, I had to add the listener to a command. I must type the command “doit”, then speak. Doesn’t seem optimal. What would be cool is to have a background thread listening for a trigger i.e. “Hey CAD, open all recent drawings”

The offline speech recognition engine ‘Sphinx’ didn’t work well at all, I’m betting the Google Speech Recognition engine would work better, though it requires configuration.

AutoCAD doesn’t understand humans, but it should be possible to build a post processor, i.e. “open all recent drawings” could be mapped to a macro that opens all the recent drawings.

What would actually be useful? I don’t think drawing geometry would be. But I could imagine.
“open drawing …”
“plot sheets […names…]”
“set UCS top”
“create new layer […name..]”

[watsonlisp]

Thanks Daniel
For my purposes i just need dictation for text prompts, not system control.
I'm working on an AI project.

[Ed Jobe]

For that, I don't see why you need AI. Just access the system's microphone and use a voice-to-text api. I don't think you will find anything in lisp though. After you have the text, then use it to set the mtext.Value property.

[watsonlisp]

no, it finds detail drawings by text and attribute content, i need speech to text for the querys
like "cleanout adapter?"
the project IS, AI.

and yes i want to spit out speech to GETSTRING or maybe a variable?

the program has a *USER:* prompt .
it would also be cool to call out commands, like lookdb, readd etc.

the programs works with text, drawings, graphics, media and html.

and thanks ed oops it should be speech recognition