Voice Notes & Speech to Text + ChatGPT

Hi everyone, I want our users to record voice notes within GlideApps for e.g. hands-free documentation and transcribe this into fine tuned and summarized notes afterwards. Any experiences with this? A simple audio recording option Glide would be fine I guess, the rest can be zapped together. Thanks you!!

1 Like

Unfortunately this is not possible as of now.

I imagine the flow would be:

  • Record audio with an app on your device
  • Upload that audio to Glide
  • With a button, send that audio to Whisper API and get the transcription
  • Use ChatGPT API to summarize that and get the text back to Glide using Glide API
2 Likes

Exactly, that would be the idea :slight_smile:

1 Like

Ahem

7 Likes

True - update came after the question I believe but I had difficulties making it work since it connects to the google cloud api. Did you?

Weā€™re unfortunately still waiting for Google to finalize the OAuth flow verification. Shouldnā€™t be much longerā€¦ We have to have it live in production to submit for verification, but nobody can use it unless its verified. Classic.

1 Like

Thanks!

Wanna build it together? I have this very functionality already built for an iOS shortcut (built it for myself). Iā€™m using OpenAIā€™s Whisper to transcribe audio though. Adapting it for Glide would be awesome :slight_smile:

Should be working now - weā€™re verified as of today. Also we added OpenAIā€™s whisper speech to text engine :smile: both work well. I prefer whisperā€™s output. Show us what you build!

5 Likes

It works! Goosebumps :smiley:

Very cool :slight_smile:

Itā€™s working for me using Whisper, however I couldnā€™t get the google version work.
According to some reviews I read it would be even a little bit better.
Iā€™m getting this error message, probably Iā€™ve don something wrong with the setup?

Itā€™s using the same audio file that is working with the openai integration.

1 Like

Hi Community and @Tim_Gestels

Trying to use OpeanAI Speech to text and receiving this error message:

Speech to text: Error 400: Bad Request - { ā€œerrorā€: { ā€œmessageā€: ā€œInvalid file format. Supported formats: [ā€˜m4aā€™, ā€˜mp3ā€™, ā€˜webmā€™, ā€˜mp4ā€™, ā€˜mpgaā€™, ā€˜wavā€™, ā€˜mpegā€™]ā€, ā€œtypeā€: ā€œinvalid_request_errorā€, ā€œparamā€: null, ā€œcodeā€: null } }

Does anyone know how to help me?

Error message image

Do you know what file extension were you using for your upload? What type of device are you trying this on?

Hi @ThinhDinh ,

mp4, according to the URL generated by the Audio Recorder from Glide. I also downloaded the file to double check and it was working fine, I mean, it was a mp4 file indeed. I was using Glide app to run/simulate the Actions.

Which OS were you trialling it on: iOS/MacOS/Android/Windows?

The Whisper API is still in beta, and it fails on mp4 files created by Safari. Users have reported this on the OpenAI forums.

2 Likes
15 Likes

This is fantastic. Another great leap for Glide.

Stepping back, I celebrate Glideā€™s new found revenue driver, which is to leverage users clamor to integrate external services. I really hope this helps minimize the cost for apps that are designed to be integration-free.

Hi @ThinhDinh,

MacOS/Safari.

I Think @david nailed. I will try to use Chrome. Hope it works.

All the best and thank you all the for the help!

Edu

@Sebastian_Daus have you compared how Whsiper performs in comparison to iOS or Androids dictation function?