Is SOUND recognition possible?

Could AutoVoice be made to recognise sounds or would OK Google etc. need to do that? I'm thinking that if a recording of a person's voice was stored recognition of a command would be more accurate. It would also make recognition language independent. Also, people with speech impediments would be able to set up commands which they could easily say. If this is something Autovoice could do and would not involve too much work, I would like to request it, please.

 

Do you want the phone to actually play the sound and then recognize it? I mean if you're going to store the voice command you'd have to have a button to access the stored sound, like a button in a scene. And in that case you could just make a menu with each button connected to a different command. If you want to record a sound to go with it just make the sound play before executing the command. If you really wanna connect it to a voice command, Autovoice has a "test command" option. You can set any text you want (including variables) and autovoice will simulate a command recognized.

 

No. Instead of interpreting a spoken word and compairing it to a stored command word I want the actual voice to be stored (in mp3 perhaps) and compare with the actual voice spoken as a command. As an extreme example someone unable to speak but can whistle or hum could use different frequencies for different commands, or someone with a lisp could say 'return' and have it recognized - 'leturn' not
being an English word. The sound would only usefully be played back so the speaker could check it had been recorded properly.You don' t need a button to recognize a command with AutoVoice and Tasker. You tell Autovoice the word which starts a particular task and when it is recognized the task is performed. I would like an AutoSound? to do exactly the same thing for a recognized sound instead of a recognized (and interpreted) spoken word. Many people have speech impediments which make them difficult to be understood by humans and impossible currently by computers.

 

Ah ok. Sorry, I misunderstood your original intentions. That's a question I don't know the answer to Hopefully Joao will be able to answer. Sorry for the misunderstanding!

 

No problem. I probably didn't explain very well.

 

Thanks for replying Joao. I know it is not currently possible which is why I raised it as an app suggestion. I appreciate that you have higher priorities. Programs such as Soundhound which can identify a song realtime from a snippet must have developed comparison of sound files so I'll ask their developers if they can produce an app which compares 2 sounds and returns a percentage match which can be accessed by program.Thanks again.

 

I might be misunderstanding the question, or since I've never used this feature maybe I misunderstand what it does...but couldn't you use the Speak Filter option in the Autovoice Recognized profile?

 

AutoVoice is
for VOICE. I. E. spoken words which are translated into written words e.g. 'help' when you say help. To recognize a whistle, say, you would need to store the text representation of that sound but it does not have one. Soundhound, for example, can recognize tunes which do not have lyrics. You would need to compare a recording of a whistle with the whistle command in order to recognize it. I hope this helps clarify.

 

Yes, it looks like I misunderstood your op...I thought you were trying to match a recording of a person's voice. Close, but not quite what you're asking for. Sorry!

 

LOL. I WAS talking mainly about a person's voice but where that person's words cannot be represented by text, so it was my fault for being unclear. SORRY!

 

thank you when you speak about soundhound i like this

 

thanks , i agreed with you bro