It has nothing to do with Bing speech recognition. Id guess your code is looping and trying to execute something in repeat when trying to speak. Or trying to speak immediately after another speak.

or you changed the settings in the iotiny configuration. And should reset

or your network is too slow and it’s cutting off

or your ARC preferences have been changed and the default should be restored

DJ,  yes other skills are in the project the code to speak is one line command per your instructions in the CHATGPT skill setup. It was my understanding that ARC was a multi threaded application.  In this instance the camera skill was operating off a EZB4, 3 servos and a speaker are connected IoTiny the servos were holding position but not moving. Nothing else is running.   I have know clue of what settings you are talking about in the iotiny.  it is not the network.   I have not changed my preferences in ARC as this was a fresh install but I will look for a place to reset them to default.

Quote:

It was my understanding that ARC was a multi threaded application.

@Dave,  No that is not what I am saying.  ARC uses skills within a project.  So each skill is its own background process that talks to the core application which is kind of acting like a traffic cop sending information to hardware and other skills.  So in this case Bingspeech takes input from the mic translates it to text,  then sends this text through ARC core engine to the ChatGPT skill then this skill sends this information to OpenAI services which in turn sends the reply back to ChatGPT then the ChatGPT skill has a one line script that calls Audio.sayEZB($OpenAI Response) that goes through the ARC core engine and send to the IoTiny hardware through a WiFi connection.

Yes there are other skills the ARC core engine is dealing with as part of it normal design and operation.  No other skill is trying to use the speaker other than ChatGPT making that call. No other INPUT's are being sent to BingSpeech or ChatGPT when this happens.  If this was a network issue it should not only happen at the start of the Audio output.   This in my opinion is not a network issue. it is also not a computer issue as I have tested this running ARC on more than capable PC's.  It is not a power problem to the IoTiny.  If anything it seems like the hardware can't process the audio input fast enough, but I would find that hard to believe.

@smiller29. OK I understand. I really don't have much more to suggest. I've never used ChatGPT. Sounds like there is a lot of back ang forth going on. It still sounds like a process is taking a bit longer to finish and cutting off your final returned audio output.

However, have you tried to open a new, clean ARC project and "only" add the speech and chat skills you're using? Then only send one command through the system at a time? Take everything else out of the equation.

Or

Use an EZB instead of a Iotiny? Can you hard wire the EZB to the computer through USB? I know you want the little Iotiny and a wireless connection but this would be for troubleshooting.

I agree it is not a network issue. If Athena is correct ARC is creating the WAV file and transmitting it to IOTiny.  Since it is TCP the entire file is sent.  If it was streamed using UDP I would suggest network issue.

@Dave, I can try that Dave but what is that going to do for me? If it works my project needs more than just these skills so I don’t know how this helps me out.

There could be a few reasons why this is happening. One possibility is that there might be a limitation or configuration setting within the BingSpeech or text-to-speech system that truncates or cuts off longer responses. This could be due to the maximum length of the audio output or other constraints.

Another potential reason could be related to the way the integration between BingSpeech and CHATGPT is implemented. It's possible that the text generated by CHATGPT is not being passed correctly or completely to the text-to-speech system, resulting in the incomplete pronunciation of the response.