Can you update the OpenAI node to include the Whisper API please.
A bit glitchy but honestly so exciting. Whisper + GPT can really open up new doorways.
It’s been pretty solid for me so far
You have my vote
I record a lot of lectures, which need to get transcribed and then processed further. So having this node would be brilliant. Is it single handedly the best transcribe engine for foreign languages on the market.
YES PLEASE! Would also be really nice if some formatting could be done by the same node to prepare audio files for transcription to avoid issues with ffmpeg via multiple nodes that we currently have to build first to prep files.
Would love this to get added
I second this but think there should be a Whisper JAX node to be able to transcribe files larger than 25mb
Perhaps could consider Deepgram for this case, as it has a quite generous $200 free credit without credit card (as of my writing): Pricing & Plans | Deepgram.
It solves few problems:
- they have OpenAI Whisper model as their speech-to-text transcription model
- it can deal with large files compared to whisper API
- it’s fast
- relatively affordable, costs around $0.2 -$0.4 every time I transcribed an 1-hour video
- it has API to use: Introduction
Btw, if want to learn more about Deepgram, do check out my blog here: How To Transcribe And Summarize Video In 10 Minutes For Your Post-Webinar Emails - Tan Yong Sheng
This was added in 0.129.0.
That’s great. How it is added, do you mean we have a “Whisper” or “Deepgram” node now?