Hi @Matthew, you can generate a node with AI to convert base64 to audio, here's a workflow remix you can clone and directly use, it converts speech base64 from openai to audio file and uploads to GCP. https://buildship.app/remix/c53aa870-6bd7-4f86-a557-5b91a5d55026
im having issues with the groq speeceh to text though, its asking for the audio file path, have no idea where to get this since the audio is coming from firestore data
You can modify the convert base64 to audio node with AI to return a path instead of url then pass the pah to the groq node, leveraging AI feature should help you overcome these challenges.
You can check if you have provide text and the base64 generated by open AI is valid, this usually occurs when the openai credits are exhausted and a blank base64 is generated.