decided to go with deepgram as speech-to-text api as they can do sync transcribing of a 1 hour podcast in about 30s. Will be cheaper than AssemblyAI which also gave me 0 free credits to test my stuff, while Deepgram gave me $200 to test with #repodcasted