I’m trying to find a way to transcribe speech to text in my app and show the user as they are speaking (as opposed to making a recording and then transcribing the full thing). It is intended for very short bits of speech (1 or 2 sentences).
I’ve created HTML and JS files to do this and have been able to embed them into my Glide app. However, the “record” button seems non-responsive. When I run these files on there own in a browser I have to give permission to use the mic. I think this is why it isn’t working in Glide. But I’m not exactly sure.
I’m not entirely sure whether I understand your problem correctly!?
Do you want to process the voice input live and not only after the audio file has been stored as a whole?
I suspect that this is not possible in Glide - although “every” other tool probably has its limits here. It’s not without reason that Siri and Co wait until you have finished typing. The request is triggered at a defined time. You would need your own application that caches every second (or faster) “from start” and sends a new request, which is then processed and displayed accordingly. The number of your updates + the number of your tokens would increase exponentially. Possible, certainly, but sensible??
Why don’t you just show a loading gif while recording - the transcription should be displayed super quickly for short content?
On an add screen or in an inline form container, the user will need to submit in order for the data to be sent to a table and therefore be useable.
On a custom screen, the user should be able to perform the recording, hit ‘Stop’ and see their transcript appear in near real-time without having to hit a submit button.
You could use the native audio recorder component and then any component that displays text (text, rich text, hint).
Yes, I’m looking to have the users words transcribed in real-time. This is something possible with many other tools (including an object native in javascript). Glide definitely seems to only support recording a full file and then processing it - I have no trouble doing that.
My question, was more about getting an embedded object to gain permissions to use the microphone. I have created an app that does what I want, it will embed in Glide, but seems restricted by permissions.
There are some services out there that offers real-time transcription, I know DeepGram is one, but not sure how you would integrate it with Glide’s audio recorder.