CODING WITH CHATGPT: Last night, I saw that OpenAI had released a new API (gpt-4o-transcribe) for audio transcription and I was eager to try it. However, I’m still a API newbie and I wasn’t sure how to implement the sample code for my own use. So…
- I gave ChatGPT the sample code from OpenAI and asked it to guide me through the usage process.
- Following ChatGPT’s instructions, I wrote a simple Python file and the transcription worked like a charm.
- However, the transcribed text was a big block of text, so I asked ChatGPT to provide code to break them up by sentences. And so it did and I updated the code.
- Voila!
This is another example of how non-coders like me can put together simple AI solutions with minimal coding knowledge. But “minimal” doesn’t mean “zero”. You must still know how to recognise basic Python code and also know what to tell the machine to do.
Side note: Unfortunately, OpenAI can only take 25MB audio file sizes at this time. So it’s back to using good old Whisper (also from OpenAI) for transcribing big audio files. Also, Whisper is free when installed locally, but using this API costs about 5 cents for transcribing about 1700 words.
