I am trying to automatically extract and utilize the results of a speech recognition transcript of a dialogue. I thought of using “$(item.transcript)” in reference to the manual, but the actual transcript results and the transcript results extracted by “$(item.transcript)” seem to be different.
Is there a variable that will extract all transcript results?
Sorry, I’m sorry, I didn’t mean to be rude. Resolved.
It was just a bad microphone.
I changed the headset and then in item.transcript
I was able to get voice recognition.
We are currently attempting to summarize chat dialogue history using GPT.
The [item.transcript.text] variable is used to return an invalid json when POSTed to GPT. However, if a text sentence assigned to the [item.transcript.text] variable is thrown as is to GPT, it will be returned normally.
Why is this?
It is probable that the punctuation in your speech-to-text (STT) output is causing issues with the JSON format that you are trying to post.
I suggest using the “escapejson” parameter to substitute the transcript in your JSON body. Here’s an example:
I fetched the data from the workflow process to GPT this time and it worked after reworking the line feed code.