Hacker News Reader

I'd like to see them put speech recognition through their LLM as a post-processing step. I find it's fairly common for whisper to make small but obvious mistakes (for example a word which is complete nonsense in the context of the sentence) which could be easily corrected for a similar sounding word that fits into the wider context of the sentence.

Is anyone doing this? Is there a reason it doesn't work as well as I'm imagining?

View on HN

Do you mean use the LLM as a post-processing step within a ChatGPT conversation? Or generally (like as part of Whisper)? If it’s the former, I’ve found that ChatGPT is good at working around transcription errors. Regarding the latter, I agree, but it wouldn’t be hard to use the GPT API for that.

Yes I mean as part of the GUI but you're right, I hadn't thought of that: maybe transcription errors don't matter if chatGPT works out that it's wrong from the context and gives a correct answer anyway.

Reply on news.ycombinator.com